Getting timed out form your backend puts the Dialer into an unrecoverable state

Similar to the internal weird state discribed in #1214 there seems to be more timeouts and reconnects happening which are affecting our users. There are numerous errors in the console that are not being surfaced to the user via the Embeddable making the system thing they are still on a call when they aren't.

Here is a screenshot from the console of one such user

So there was timeout which then failed to disconnect the call... but then the call was actually gone and not able to be reconnected too and the user is stuck and unable to make any more calls without fully restarting their browser.

A few more errors that were seen later in that session as the user tried different things.

The websocket state seems also like y'all might be having network issues which the embeddable is having problems with? (NOTE: I've realized there is a chance these network troubles might be local too and we are investigating that rather than assume it is on your end)

Version 3.0 beta

The report concerns the Embeddable Dialer’s timeout, disconnect, reconnect, and websocket handling in Version 3.0 beta; no source file or test is named. Begin by reproducing the timeout and reconnect sequence, then trace the call and websocket state transitions. Done means the Embeddable surfaces the failure and returns to a usable calling state without a browser restart.