It replaces face-api.

* feat(face-landmarks): integrate human library

* feat(face-landmarks): rewrite worker in typescript

* fix(face-landmarks): allow worker bundle size up to 2 mib

* fix: remove unwanted comment

* code review

- opens as a modal
- lastn is 0, mutes local video while open
- long press to talk
- and more

This way only one will be shown at a time.

If we have enough space on the last row we expand the width of the tiles
up to 16:9 ratio.

It now lives here https://github.com/jitsi/jitsi-meet-load-test

* Refactor load-test into an object.

* Convert to class syntax.

* Bind member function callbacks.

* More binding and thisage.

* More thisage.

* More tweaks

* Rename numParticipants as remoteParticipants.

* Change back.

* Fix userLeft.

* Add members for event listeners, to be able to remove them.

* Add numClients parameter that allows multiple clients to be started.

* Clear clients array on unload.

* Add latency between starting clients.

* fix large video updating loop when kicked out

* fix linter issue

https://github.com/jitsi/lib-jitsi-meet/compare/v1430.0.0+ccf9ebed...v1431.0.0+742232c9

Use something less dull.

Original: https://unsplash.com/photos/Q1p7bh3SHj8

Also, stop bundling WebRTC, it's now a dependency in CocoaPods.

* Update cloud-api.swagger

Update to match with the current implementation

* Update cloud-api.swagger

Only havee the screen in the hierarchy if we have a welcome page, since
it's the only way to access it.

Use goBack() from the navigator directly and avoid duplicating all props
to the screen.

It's about to become very confusing, since we are going to add actual
local recordings with video.

This feature was never fully finalizeed since it required manual
processing of the files, as they were not uploaded anywhere.

In addition, unless one opens the local audio device without any audio
processing first, any tracks opened later will have audio proceessing
turned on, something not desirable for the scenario this feature was
designed for in the first place: podcasts.

This feature will likely come back as a JaaS demo / MVP where the local
recording is made outside of the Jitsi Meet iframe.

Just like the web.

Now that screenshare is permitted when a user is in audio-only mode, do not create a new video track if it doesn't exist when audio-only mode is automatically disabled. New tracks should only be prompted by user action such a camera unmute or start screenshare. Fixes https://github.com/jitsi/jitsi-meet/issues/11460.

https://github.com/jitsi/lib-jitsi-meet/compare/v1429.0.0+7251f3aa...v1430.0.0+ccf9ebed

It allows us to place the Cancel button there, and declutter the UI of
buttons.

Skip the welcome page for now, until we can remove the audio / video
toggle.

* fix: Fixes loading recommendedBrowsers page.

Fixes loading the page and its resources when using html base (cdn).

https://github.com/jitsi/lib-jitsi-meet/compare/v1428.0.0+eda20c2a...v1429.0.0+7251f3aa

* fix(multi-stream) update selector to find ss track by videoType or mediaType

* ref(multi-stream) move fake ss creation logic and support video type changed

* refactor(multi-stream) decouple sending and receiving multiple screenshare streams

* fix(multi-stream) fix receiver constraints with signaling and without multi-stream

* fix(mutli-stream) ensure plan b original SS thumbnail displays avatar

* fix(multi-stream) show fake SS for plan b sender

* refactor(multi-stream) poc for moving SS creation to state listener

* remove reference to fake SS creation

* fix lint errors

* rename to virtual screenshare participants

* fix minor bugs

* rename participant subscriber to specify web support only

Since all the auth logic is not ported to React on the web,
`_isWaitingForOwner` will always return `false` because the
`waitForOwner()` action is not (yet) used there.

THis fix always tries to hide the dialog no matter what, which is not a
Bad Thing to do anyway.

There is a related bug remaining, however: if one pressed "I am the
host" and then cancel, it doesn't goo back to the previous dialog, but
it completely kils the meeting. This is a compromise we'll have to live
with for a bit longer.

Fixes: https://github.com/jitsi/jitsi-meet/issues/11413
Fixes: https://github.com/jitsi/jitsi-meet/issues/11432

* hideAutoAssignButton
* hideFooterMenu
* hideModeratorSettingsTab
* hideMoreActionsButton
* hideMuteAllButton

If downloading the model tiemouts, it may succeed in the (near) future,
don't just give up.

* fix occasional unexpected avatar

* fix ss stats popover not shown

* fix linter issue

https://github.com/jitsi/lib-jitsi-meet/compare/v1425.0.0+6b629a19...v1428.0.0+eda20c2a

We used to multiplex the ports in nginx, but we dropped that at some point, so now coturn is on its own listening and nginx dependency is no longer needed. Our turnserver config can be used with nginx | apache2.