This PR takes The Bulldozer Approach (R): removes the old toolbox and lots of
associated code, though not all of it. Subsequent cleanups will follow.

* feat(recording): use google api to get stream key

* squash: renaming pass

* squash: return full load promise

* sqush: use google api state enum

* squash: workaround for lib not loading

* another new design...

* increase timeout workaround for gapi load issue

* styling pass

* tweak copy

* squash: auto select first broadcast

Adds base/sounds feature which allows other features to register a sound
source under specified id. A new SoundsCollection component will then
render corresponding HTMLAudioElement for each such sound. Once "setRef"
callback is called by the HTMLAudioElement, this element will be added
to the Redux store. When that happens sound can be played through the
new 'playSound' action which will call play() method on the stored
HTMLAudioElement instance.

The animation for toggling filmstrip visibility was lagging on
Safari. Even though the background video is set to hidden, it is
still causing issues. Setting the background to display none
instead does help but might interfere with animations. So instead
do the easy thing and re-use logic used for Firefox to not show
the background video.

* deps: update jquery version

* squash: resize thumbnails after appending shared thumb

This forces jquery animate to show the thumbnail somehow...
Remote thumbnails basically work this way (append to filmstrip
and then resize filmstrip thumbnails) so I just copied that
implementation. ... So I admit I lost this fight because
even after looking at jquery I couldn't understand why
it doesn't work on the first resize but does on the second.
Plus I'm being put on a strict timebox to update jquery.

* squash: getJSON no longer supports .success

The difference from this change and 88325ae is there is no
attempt to do this in redux. This is the safer change in that
the cleanup logic is known only to trigger on hangup.

This reverts commit 88325aeef2.
Turns out a conference with a password triggers a failed conference
join. It's going to be tricky to decipher when to do actual
cleanup, and where to shove that code, so reverting is easier for
now.

Destroy local tracks and also destroy large video so the
user does not wonder why camera (and mic) are still enabled
even though hangup has been pressed.

* feat(welcome): new design

* squash: update strings

* squash: copy/paste error?

* squash: remove welcome page disabling checks

* squash: change strings again

* squash: background var

* squash: title and desc css as variables

* ref(large-video): reactify background

This is pre-requisite work for disabling the background on
certain browsers, namely Firefox. By moving the component
to react, and in general encapsulating background logic,
selectively disabling the background will be easier.

The component was left for LargeVideo to update so it can
continue to coordinate update timing with the actual large
video display. If the background were moved completely into
react and redux with LargeVideo, then background updates would
occur before large video updates causing visual jank.

* fix(large-video): do not show background for Firefox and temasys

Firefox has performance issues with adding filter effects on
animated elements. On temasys, the background videos weren't
really displaying anyway.

* some props refactoring

Instead of passing in classes to LargeVideoBackground, rely on
explicit props. At some point LargeVideo will have to be reactified
and the relationsihp between it and LargeVideoBackground might
change, so for now make use of props to be explicit about
how LargeVideoBackground can be modified.

Also, set the jitsiTrack to display on LargeVideoBackground to
null if the background is not displayed. This was an existing
optimization, although previously done with pausing and playing.

* squash: use newly exposed RTCBrowserType

* squash: rebase and use new lib browser util

* squash: move hiding logic all into LargeVideo

* squash: remove hiding of background on stream change. hopefully doesnt break anything

Passing around of the component was used when there were two
independent Notification components. Now that there is only
one Notification component, it is not necessary to pass
around the component.

Instead of having visible and visibility and setToolboxVisible and
setFilmstripVisibility, have only visible as a name.

The current notification for starting muted is 2 minutes, which
may seem like "forever" so reduce it to dismiss faster.

Spot will need a way to submit call feedback using the iframe
api. For now expose a method on conference.js to submit that
feedback. Exposing on conference.js looks to be the existing
pattern... Also add an event to notify consumers of the iframe
api that feedback was submitted, as postMessage is async
and the notification can at least give some guarantee maybe.

I haven't updated documentation yet as I'm not confident
about this api.

* ref: Restructures the pinned/unpinned events.

* ref: Refactors the "audio only disabled" event.

* ref: Refactors the "stream switch delay" event.

* ref: Refactors the "select participant failed" event.

* ref: Refactors the "initially muted" events.

* ref: Refactors the screen sharing started/stopped events.

* ref: Restructures the "device list changed" events.

* ref: Restructures the "shared video" events.

* ref: Restructures the "start muted" events.

* ref: Restructures the "start audio only" event.

* ref: Restructures the "sync track state" event.

* ref: Restructures the "callkit" events.

* ref: Restructures the "replace track".

* ref: Restructures keyboard shortcuts events.

* ref: Restructures most of the toolbar events.

* ref: Refactors the API events.

* ref: Restructures the video quality, profile button and invite dialog events.

* ref: Refactors the "device changed" events.

* ref: Refactors the page reload event.

* ref: Removes an unused function.

* ref: Removes a method which is needlessly exposed under a different name.

* ref: Refactors the events from the remote video menu.

* ref: Refactors the events from the profile pane.

* ref: Restructures the recording-related events.

Removes events fired when recording with something other than jibri
(which isn't currently supported anyway).

* ref: Cleans up AnalyticsEvents.js.

* ref: Removes an unused function and adds documentation.

* feat: Adds events for all API calls.

* fix: Addresses feedback.

* fix: Brings back mistakenly removed code.

* fix: Simplifies code and fixes a bug in toggleFilmstrip

when the 'visible' parameter is defined.

* feat: Removes the resolution change application log.

* ref: Uses consistent naming for events' attributes.

Uses "_" as a separator instead of camel case or ".".

* ref: Don't add the user agent and conference name

as permanent properties. The library does this on its own now.

* ref: Adapts the GA handler to changes in lib-jitsi-meet.

* ref: Removes unused fields from the analytics handler initializaiton.

* ref: Renames the google analytics file and add docs.

* fix: Fixes the push-to-talk events and logs.

* npm: Updates lib-jitsi-meet to 515374c8d3.

* fix: Fixes a recently introduced bug in the google analytics handler.

* ref: Uses "value" instead of "delay" since this is friendlier to GA.

Shared video (youtube videos) do not have indicator icons about
dominant speaker and raised hand. Add a check to see if the
indicator icon container exists before trying to attach react
icons to it. The error thrown didn't seem to be causing any
UI issues though.

While this truthy check before doing ReactDOM.render seems
like it should be a general practice, I'm hesistatnt to make
the change broadly because it hasn't been a problem so far.

* ref(avatars): remove Avatar.js

- Rely on redux getting updated with new participant state
and any calls to getAvatarURL passing in the redux
participant state. This way the state within Avatar.js can
be removed.
- Clean up methods on UI.js. Because all state is in the
store, separate methods for updating the avatar aren't as
necessary. Instead centralize accessing of the avatar for
components outside of redux and centralize the call to
update avatars for non-react components.
- Controversial: cache a participant's avatarURL on the
participant state. Currently the participant's avatarURL
that is generated without jwt (which sets the avatarURL directly)
is not cached. Without cache, there can be many redundant
calls to APP.API.notifyAvatarChanged.

* Leverage middleware timing to diff avatars

One alternative implementation is to leverage middleware's
ability to intercept updates before and after redux has
upated and then compare avatarURLs.

* kill UI.getAvatarUrl

* profile button sets its own avatar url (solves update timing)

* remove calls to updating avatar outside of middleware

* update UI.js doc

* remove left over logic from initial implementation

* try to move local user fallback into selector func

* default to id 'local' in selector

* ref(analytics): centralize all event names

* squash: fix typo and alpha ordering

* squash: rename file to AnalyticsEvents to parallel lib-jitsi-meet

- Modify Etherpad and SharedVideo so their resizing takes into account
  the width of the filmstrip in vertical filmstrip mode.
- Modify Filmstrip's getFilmstripWidth to account for when the filmstrip
  is hidden.
- modify VideoContainer so in vertical filmstrip mode it centers the
  shared desktop stream in the middle of the available space not taken
  by filmstrip.
- Also allow clickthrough on the secondary toolbar itself while still
  allowing clicks on the toolbar's buttons. This allows clicks on
  shared videos to go through.

Using column-reverse prevents proper scrolling on browsers other than Safari
and Chrome. Additionally, Firefox has an issue where flex containers have
dimensions set to auto, preventing resize. So, add hacks to maintain Chrome
and Safari's behavior while allowing for some kind of scrolling on other
browsers.

The goal is to reduce usage on atlassian/aui. New components
have been created to display the settings panel. Language
selection will reach into i18n for state whereas moderator
options will keep state in redux.

* On adding participant show the initial status text if any.

When participant joins and there is an initial status we show it, instead of the connected notification. If initial status is missing proceed with the default user connected notification.

* Updates to latest lib-jitsi-meet (365bd886ee).

* Adds a fixme comment.

* Updates to latest lib-jitsi-meet (365bd886ee).

* Adds initial documentation for sipgw jibri.

Also explains enabling the people search service and the request/response that are made around sipgw jibri service.

* Fixes add people dialog to invite users and rooms.

No invitation is sent when there is nobody to invite.

* Reuse some recording strings, by using arguments.

* Make sure web also dispatches CONFERENCE_WILL_JOIN.

* Introduces new feature videosipgw.

* Fixes lint errors.

* Renames methods to use people, chatRooms and videoRooms.

* Updates to latest lib-jitsi-meet (dc3397b18b).

On DOMINANT_SPEAKER_CHANGED events, switch to the new
dominant speaker regardless of if the speaker has video;
it's okay to show the speaker's avatar.

Always switch to the dominant speaker after unpinning
instead of conditionally checking if the current dominant
speaker has video.

It was decided along with the mute participant dialog reactification
that these types of warning messages should not be toggleable--that
they should simply always display because there is no undo action.
As such, the component NotificationWithToggle is no longer needed.

* fix(notifications): throttle and batch join notifications

Instead of directly calling to show a join notification,
go through a specific method. This method will queue
names for display while a throttled function pulls
the names and shows a notification.

* squash: remove unused translation key

* squash: use default display name

* squash: move into participant actions

After looking at the jquery-ui documentation, I believe it
is being used only in one place, when toggling the smiley
menu. That toggling has been quickly replaced with a normal
jquery call.

Note: @atlassian/aui also uses jquery-ui but killing that
off will be more difficult due to its CSS being used.

Prepares for the latest eslint & flow-related npm packages which cause
these errors.

Click handling was added to the local thumbnail only after
receiving video. Click handling was added to remote thumbnails
only after receiving a stream. To allow for pinning through
the filmstrip in any case, always attach the click handler
on thumbnail init.

Checks exist when clicking a contact list to prevent the pinning
UI from updating if a remote thumbnail has not loaded video.
This was unexpected UX so instead go ahead and pin. This is
accomplished by having contact list clicks to more directly
trigger the flow of a thumbnail being clicked.

The YouTube api removed player#getVideoData years ago.
I am completely baffled as to how the feature kept working.
The title of the video is no longer obtainable through
the player itself, so instead call the bot participant
"YouTube."

* feat(keyboard-shortcuts): show help in a react dialog

- Move shortcut help dom declaration to a react component
- Let open/close logic be handled by AtlasKit Dialog
- Remove static html for help from index.html
- Consolidate keyboard shortcut css

* squash: use lozenge for key styling

* squash: use different iteration style

* squash: update package-lock for lozenge