Fix exceptions when handling incoming transactions

 Add a five minute cache to get_destination_retry_timings

Rewrite the dockerfile as a multistage build: this means we can get rid of a whole load of cruft which we don't need.

remove spurious federation checks on localhost

Remove redundant call to start_get_pdu_cache

I think this got forgotten in #3932. We were getting away with it because it
was the last call in this function.

Improve performance of getting typing updates for replication

Build and push docker image to hub automatically

Clarifications in FederationHandler

Run notify_app_services as a bg process

Replaced all occurences of e.message with str(e)

This test didn't do what it claimed to do, and what it claimed to do was the
same as test_cant_hide_direct_ancestors anyway.

This stuff is tested by sytest anyway.

Signed-off-by: Schnuffle <schnuffle@github.com>

fix #3854 MAU transaction errors

Latest is horrible and makes debugging what has happened anywhere a
nightmare. We push a latest because of demand for it, but we'll also
push a SHA1 commit id so those wanting to know what they're running
(and be able to roll back if required) can use those instead.

Note that latest here is defined as "most recent master commit" not
"most recent released version", as the actual semantics of making latest
correct while still being able to build bugfixed releases of previous
versions is just ARGH. So we define it as "master" not "latest release".

If we have a forward extremity for a room as `E`, and you receive `A`, `B`,
s.t. `A -> B -> E`, and `B` also points to an unknown event `X`, then we need
to do state res between `X` and `E`.

When that happens, we need to make sure we include `X` in the state that goes
into the state res alg.

Fixes #3934.

If we've fetched state events from remote servers in order to resolve the state
for a new event, we need to actually pass those events into
resolve_events_with_factory (so that it can do the state res) and then persist
the ones we need - otherwise other bits of the codebase get confused about why
we have state groups pointing to non-existent events.

get_state_groups returns a map from state_group_id to a list of FrozenEvents,
so was very much the wrong thing to be putting as one of the entries in the
list passed to resolve_events_with_factory (which expects maps from
(event_type, state_key) to event id).

We actually want get_state_groups_ids().values() rather than
get_state_groups().

This fixes the main problem in #3923, but there are other problems with this
bit of code which get discovered once you do so.

* add some comments on things that look a bit bogus
* rename this `state` variable to avoid confusion with the `state` used
  elsewhere in this function. (There was no actual conflict, but it was
  a confusing bit of spaghetti.)

Logging improvements