crypto: Fetch immediately-available sender data when we receive a room key

[andybalaam]

Part of #3543. Builds on top of #3556

Implements the "fast lane" as described in #3544

This will begin to populate InboundGroupSessions with the new SenderData struct introduced in #3556 but it will only do it when the information is already available in the store. Future PRs for this issue will query Matrix APIs using spawned async tasks.

Future issues will do retries and migration of old sessions.

[andybalaam]

Rebasing because this is a mess after the merge of #3556

[andybalaam]

Rebased on top of #3633

[codecov]

Codecov Report

Attention: Patch coverage is 86.58537% with 11 lines in your changes missing coverage. Please review.

Project coverage is 84.37%. Comparing base (a5dbfa6) to head (d496196).
Report is 18 commits behind head on main.

Files	Patch %	Lines
...rypto/src/olm/group_sessions/sender_data_finder.rs	80.48%	8 Missing ⚠️
...k-crypto/src/session_manager/group_sessions/mod.rs	87.50%	3 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #3590   +/-   ##
=======================================
  Coverage   84.37%   84.37%           
=======================================
  Files         258      259    +1     
  Lines       26656    26722   +66     
=======================================
+ Hits        22491    22547   +56     
- Misses       4165     4175   +10     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[andybalaam]

@poljar this is now rebased and ready to review.

[poljar]

I think we might simplify this a bit. Please correct me if I'm wrong.

[poljar]

I think that this can be further simplified if we use the functionality the Device offers.

[poljar]

Alright, I think this is functionally fine, except maybe one thing I raised we should discuss.

I left a bunch of documentation nits, feel free to fight back on them if you disagree.

[poljar]

I think that this is good now, let's just clear up the history I wonder why our CI check for fixup commits didn't kick in, perhaps because the PR is older than the CI check?

[andybalaam]

I wonder why our CI check for fixup commits didn't kick in

Or maybe because I just manually named the commits with fixup: rather than using --fixup because that doesn't allow me to describe the commit (unless I missed a flag)