fix(base): clear a room's send queue and dependent event queue after removing it from the state store

[bnjbvr]

Since this wasn't properly done for the dependent send queue for rooms, technically we should run a migration to look for the entries in the dependent send queue table, and remove all those that don't correspond to known room ids. I didn't bother here because the value seems quite low (this shouldn't affect many lines), but if the reviewer strongly feels about it, we could clean it up.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 85.02%. Comparing base (0b16d48) to head (56863b4).
Report is 4 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4295      +/-   ##
==========================================
- Coverage   85.02%   85.02%   -0.01%     
==========================================
  Files         274      274              
  Lines       30107    30126      +19     
==========================================
+ Hits        25599    25615      +16     
- Misses       4508     4511       +3     

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

---

🚨 Try these New Features:

• Flaky Tests Detection - Detect and resolve failed and flaky tests

[poljar]

Don't we need this for the IndexedDB store? Don't we have support for the send queue there?

[bnjbvr]

Thanks for the review!

It was correctly implemented for the indexeddb state store 🙃 (as proven by the tests not failing without any additional changes there). So yeah, our implementations were a bit inconsistent before…