They're useless most of the time and seem to delay Firefox/Chrome when
trying to resolve them.

See
  https://chromium.googlesource.com/external/webrtc/+/194e3bcc53ffa3e98045934377726cb25d7579d2/webrtc/media/engine/webrtcvoiceengine.cc#302
  https://bugs.chromium.org/p/webrtc/issues/detail?id=8133

Technically it's against the spec to modify the SDP but there's no other
API for this and this seems to be the only possible workaround at this
time.

Work around negotiation problem in jackaudiosrc that causes it to negotiate stereo even if mono is configured

All exceptions are logged and cause the current stream to be stopped.

Otherwise there will be a 5s delay until the table is refreshed before
our own listener is counted in the numbers.

Closing the websocket connection is sufficent for deleting/leaving
rooms.

Not actually needed anywhere.

Lists all currently available rooms and has some play/pause logic.
WebRTC integration is still missing.

Add commandline parameter for providing the server certificate.

Does not implement any configuration yet and simply streams an
audiotestsrc.

We can just called shutdown() from stopped(), it does exactly the same.

Otherwise we create reference cycles.

No need for another round-trip via messages to get it from the room.