Key requests wait for a short time (500ms) before being sent as
an attempt to wait for the key to arrive before sending a request
for it, but client startup takes way longer than this so this
would still result in key requests being sent for keys that we'd
fetched but were still waiting to be read out of the sync response
and put into the database.
The bulk of this is just export/import changes, though there's a couple pieces to highlight:
* We no longer use default exports. This is because it's discouraged by the JS community, though not in any official capacity.
* We now use `polyfillSuper` for some prototype inheritance because the tests, and sometimes webpack, break on "cannot call EncryptionAlgorithm without 'new'". It's very much a workaround, and definitely not needed when we use real classes.
There is some import shuffling to help keep the imports clean - this was done by my IDE.
use sendRoomKeyRequest with a new resend flag, instead of cancelRoomKeyRequest,
when requesting keys, so that we make sure that we send a new request if there
is no previous request
fixes https://github.com/vector-im/riot-web/issues/6838
* Start a new Olm sessions with a device when we get an undecryptable
message on it.
* Send a dummy message on that sessions such that the other end knows
about it.
* Re-send any outstanding keyshare requests for that device.
Also includes a unit test for megolm that isn't very related but came
out as a result anyway.
Includes https://github.com/matrix-org/matrix-js-sdk/pull/776
Fixes https://github.com/vector-im/riot-web/issues/3822
Hopefully this will fix the vector-web test failures (the
OutgoingRoomRequestManager throws an exception because the indexeddb is being
deleted just as it's getting started).
