When a client asks for users whose devices have changed since a token we
used to pull *all* users from the database since the token, which could
easily be thousands of rows for old tokens.

This PR changes this to only check for changes for users the client is
actually interested in.

Fixes #5553

* master:
  Fix broken link in MSC1711 FAQ
  Update changelog to better expain password reset change (#5545)

* release-v1.0.0:
  Update changelog to better expain password reset change (#5545)

Closes #4583

Does slightly less than #5045, which prevented a room from being upgraded multiple times, one after another. This PR still allows that, but just prevents two from happening at the same time.

Mostly just to mitigate the fact that servers are slow and it can take a moment for the room upgrade to actually complete. We don't want people sending another request to upgrade the room when really they just thought the first didn't go through.

Fixes #3370.

Python 3.7 is apparently faster than 3.6, and should be mature enough.

Updates the v1.0.0 changelog to provide more information to those upgrading to v1.0 on why they may receive errors or find their password reset abilities have now been disabled.

Helps address #5444

Fixes #5533

Adds information about how to install and run black on the codebase.

Add --data-dir and --open-private-ports options.

Stop conflating generated config and default config

Cleanups and sanity-checking in cpu and db metrics

Split public rooms directory auth config in two

Fix /messages on workers when no from param specified.

This is helpful when generating a config file for running synapse under docker.

We don't necessarily want to put the data in the cwd.

It's too confusing.

This will enable us to skip the unintuitive behaviour where the generated
config and default config are the same thing.

Allow configuration of the path used for ACME account keys.

This is no longer used and only serves to confuse.

Because sticking it in the same place as the config isn't necessarily the right
thing to do.

* Pull config_dir_path and data_dir_path calculation out of read_config_files

* Pass config_dir_path and data_dir_path into read_config

This makes some of the conditional-import hoop-jumping easier.

If no `from` param is specified we calculate and use the "current
token" that inlcuded typing, presence, etc. These are unused during
pagination and are not available on workers, so we simply don't
calculate them.

Sentry will catch the errors if they happen, so that should be good enough, and
woun't make things explode if we hit the error condition.

Use monotonic clock where possible for metrics

This has no useful purpose on python3, and is generally a source of confusion.