Fix errors storing large retry intervals.

Fix fetching censored redactions from DB

We have set the max retry interval to a value larger than a postgres or
sqlite int can hold, which caused exceptions when updating the
destinations table.

To fix postgres we need to change the column to a bigint, and for sqlite
we lower the max interval to 2**62 (which is still incredibly long).

Fix censoring redactions performance

Fetching a censored redactions caused an exception due to the code
expecting redactions to have a `redact` key, which redacted redactions
don't have.

Joining against `events` and ordering by `stream_ordering` is
inefficient as it forced scanning the entirety of the redactions table.

This isn't the case if we use `redactions.received_ts` column as we can
then use an index.

This will allow us to efficiently search for uncensored redactions in
the DB before a given time.

Currently we don't set `have_censored` column if we don't have the
target event of a redaction, which means we repeatedly attempt to censor
the same non-existant event.

When we persist non-redacted events we unset the `have_censored` column
for any redactions that target said event.

Fix `devices_last_seen` background update.

Fixes #6134.

Fix 'redaction_retention_period' sampel config to match guidelines

formatting tweak

Fixes #5905

Prune old rows in user_ips tables.

Move last seen info into devices table

Enable cleaning up extremities with dummy events by default to prevent undue build up of forward extremities. (#5884)

Co-Authored-By: Richard van der Hoff <1389908+richvdh@users.noreply.github.com>

Fixes #6103

If email or msisdn verification aren't supported, let's stop advertising them
for registration.

Fixes #6100.

Pull the checkers out to their own classes, rather than having them lost in a
massive 1000-line class which does everything.

This is also preparation for some more intelligent advertising of flows, as per #6100

because, frankly, it looked like it was written by an axe-murderer.

This should be a non-functional change, except that where `m.login.dummy` was
previously advertised *before* `m.login.terms`, it will now be advertised
afterwards. AFAICT that should have no effect, and will be more consistent with
the flows that involve passing a 3pid.

Fix the return value in the users_set_deactivated_flag background job