Fix serialisation errors when using third-party event rules.

* Tie together matches_user_in_member_list and get_users_in_room

* changelog

* Remove type to fix mypy

* Add `on_invalidate` to the function signature in the hopes that may make things work well

* Remove **kwargs

* Update 8676.bugfix

* Fix user_daily_visits to not have duplicate rows for UA.

Fixes #8641.

* Newsfile

* Fix typo.

Co-authored-by: Patrick Cloke <clokep@users.noreply.github.com>

don't bother constricting a CacheContext unless we need one.

Optimisation in DeferredCache.set

Push some deferred wrangling down into DeferredCache

* Limit AS transactions to 100 events

* Update changelog.d/8606.feature

Co-authored-by: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com>

* Add tests

* Update synapse/appservice/scheduler.py

Co-authored-by: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com>

This should (theoretically) allow for using the TCP code with a different output type
and make it easier to use the JSON code with files / console.

I noticed in https://github.com/matrix-org/synapse/issues/8575 that the `end_error` variable in `synapse_port_db` is set to an `Exception`, even though later we expect it to be a `str`.

This PR simply casts an exception raised to a string. I'm doing this instead of having `end_error` be of type exception as we explicitly set `end_error` to a str here:

https://github.com/matrix-org/synapse/blob/d25eb8f3709965d0face01a041d5292490bf0139/scripts/synapse_port_db#L542-L547

This whole file could probably use some heavy refactoring, but until then at least this fix will prevent exception contents from being hidden from us and users.

* Move schema file

* Add a .

* Add matching changelog entry

* Fix sqlite

We need to make sure we are readu for the `set_cache_factor` callback.

Signed-off-by: Vasilis Gerakaris <vasilis.gerakaris@navarino.gr>

* Add `DeferredCache.get_immediate` method

A bunch of things that are currently calling `DeferredCache.get` are only
really interested in the result if it's completed. We can optimise and simplify
this case.

* Remove unused 'default' parameter to DeferredCache.get()

* another get_immediate instance

By using the "poll" reactor since macOS doesn't support epoll.

This should reduce the number of `There was no active span` errors we
see.

Fixes #8510.

Most of these uses don't need a full-blown DeferredCache; LruCache is lighter and more appropriate.