Emit ACTION_JINGLE_SI_TIMEOUT analytics event if it takes too long for
Jicofo to send session-initiate (it usually arrives within 1 second).

JingleSessionPC cleanup

The previous check looks like it was meant to check if it's
the initiator.

The options are coming from the config anyway.

* feat: Adds the user's region to presence. Fixes typos.

* feat: Measures end-to-end RTT with ping over the data channel.

* feat: Adds a couple of permanent analytics properties.

feat(RTC): refactor logic for polling for device changes

- Don't use it on supported browsers
- Make sure no stray timers are left

fix(audio-output): return the reject promise

* Enables adapter for edge.

We were not filtering correctly all unsupported iceServers and an error is thrown and no connection is established.

* Enable desktop sharing for Edge.

Currently when replacing video track with desktop sharing one, doesn't work because of: https://developer.microsoft.com/en-us/microsoft-edge/platform/issues/17528460/
If Edge user joins first and enables desktop sharing it will work when others join. Tried also to use replaceTrack/setTrack as a workaround, but again we hit an error, this time InvalidAccessError.

* Adds the helper function usesAdapter in BrowserCaps.

* feat(conference): expose method to select multiple participants

* squash: fix typo

* squash: reduce to one code flow

* squash: default to empty array to mimic previous null default

* squash: fix typos, change docs, default params

* squash: remove defaults

* squash: update comment about array arg

* ref(video-quality): cache max frame height and send on channel open

Currently it is possible to try to change the max receive video
frame height before the data channel is open. In that case an error
will be thrown. This change makes it so that the desired frame height
is saved and sent on channel open, avoiding the thrown error the
max receive video frame height logic is exercised through the
JitsiConference api.

* squash: do the second part of the actual fix

* Change layer suspension to use parameters in RTPSender

We no longer suspend unused simulcast layers via a bandwidth cap in the
SDP, instead we'll use the new parameters in RTPSender to enable and
disable streams explicitly.  The main advantage here is the RTPSender
method ramps up immediately when we re-enable the layers (as opposed to
the SDP bandwidth cap which took 30+ seconds).

* Fix linter issues

* Uses json message extension

* Sends json-message without the body element

* Info comments and parameter order changes

1. The example was using the Promise return value of JitsiMeetJS.init
which is no longer possible/correct after commit "core: refactor
initialization not to return a Promise".

2. We went back and forth with the value returned by JitsiMeetJS.init:
we initially didn't return a value, then we started returning a Promise,
and now we're not returning a value. Whether we'll go back to returning
a value is up in the air. Anyway, the return value is practically
determined by the last in a chain of function calls: JitsiMeetJS, RTC,
RTCUtil. Since the chain is not really documented, it will not hurt much
to make it easier to refactor the chain by "composing" the functions.

There is nothing asynchronous about the initialization process (anymore), thus
turn it into a synchronous method.

In addition, WebRTC support is absolute, it cannot change from not being
supported to being supported (as it plreviously could, thanks to Temasys) so get
rid of the ancillary logic to support that.

Last, introduce a way to check if WebRTC is supported in the current
environment: JitsiMeetJS.isWebRtcSupported().

RTCUtils.attachMediaStream was changed not to return elements;
instead it returns undefined by default. When mapping over
containers and call RTCUtils.attachMediaStream, containers
would be changed to undefined.

The case where we had created recvonly streams (start muted on FF) and we are unmuting, this creates a sendrecv stream and adds msid, we need to signal msid so listeners to be notified and create appropriate audio/video element and to start receiving the stream.

User may decide to abort the connection while it's stuck at connecting.

* Update reference to Prosody bugtracker.

* Updated refernce to Prosody module.

* Remove unused child element.

The child element in the query suggests that a specific host is being requested. Neither
the Prosody implementation nor the XEP-0215 specification defines this element. Its
inclusion is ignored by the XEP-0215 server implementation.

With the update of react-native-webrtc to M67, the ability to stop the camera
when a track is disabled was introduced, so there is no longer a need for doing
a local O/A to remove the track for video muting.

When receiving presence, the XML is converted to JSON. In
the process, Strophe.getText is used, which calls its
xmlescape utility function. This is not desired as the
raw value is desired, especially for display name.

Note: this issue only affects presence as it is the only
place that call the helper packet2JSON, which is the only
place that calls Strophe.getText.

* ref(recording): change implementation to match VideoSIPGW

VideoSIPGW takes in a chat room and uses instance variables
on the chat room. RecordingManager has been changed to
mirror this approach because of the case where jitsi is
deployed on a domain requiring authentication. In that
case, the initial chat room is created and fails, a
new chat room (connection) is created for authentication,
authentication is put onto the failed chat room, and the
failed chat room is used. As such, recordingManager
should not be tied to chat room creation itself but
rather tied to the first chat room.

* squash: use git mv to detect capitalization change

the docs denote this behavior, but it appears to have been removed at
some point

feat(JitsiConference): add 'getConnection' getter

* Leave jvb connection only with turns server.

* Adds a comment.

* fix(migration): reject the data channel to restart it

Starting on chrome 63, an error occurs when renegotiating
with the new bridge on bridge migration when the SDP sections
do not match, which occurs when removing the data channel
section. Not doing the double renegotiate prevents the data
channel from being reopened. The way around this, while still
avoing the error, is to reject the data channel section instead
of removing it to trigger a new data channel.

* squash: reword comments

* change connection quality to take layer suspension into account

* Support layer suspension

Add support for a message which notifies the endpoint whether or not it
is selected (meaning its HD stream is in use).  If it is not
selected and enableLayerSuspension is set to true then it will impose a
bandwidth limit in the SDP to suspend sending the higher layers.

* only add the IS_SELECTED_CHANGED listener in JingleSessionPC if layer
suspension is enabled

this prevents doing a local o/a when we don't need it