feat(event cache): include read receipts that were dormant in the state store

[bnjbvr]

When computing unread counts, if there was no read receipt marked as latest_active, it could be that a read receipt was dormant in the state store, before the code was migrated. In this case, we wouldn't take it into account for the current room, and this would result in an incorrect unread count for that room, until a new receipt is sent for that room.

I think this fixes #6211, as this situation would likely arise for a user with sync v2, migrating to the new SDK where the event cache handles read receipts.

• I've documented the public API Changes in the appropriate CHANGELOG.md files.
• This PR was made with the help of AI autocomplete.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.86%. Comparing base (32e9626) to head (3063a11).
⚠️ Report is 6 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #6406      +/-   ##
==========================================
+ Coverage   89.85%   89.86%   +0.01%     
==========================================
  Files         378      378              
  Lines      103384   103372      -12     
  Branches   103384   103372      -12     
==========================================
+ Hits        92893    92895       +2     
+ Misses       6928     6910      -18     
- Partials     3563     3567       +4     

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

[codspeed-hq]

Merging this PR will improve performance by ×2.5

⚡ 1 improved benchmark
✅ 49 untouched benchmarks

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	Restore session [memory store]	276.5 ms	112.3 ms	×2.5

---

_{Comparing bnjbvr/handle-dormant-read-receipts (3063a11) with main (a1157d2)}

[poljar]

Looking good, thanks.

[poljar]

It might make sense to put this array into a global const since it's now used here as well as in select_best_receipt().

While it's not very likely that we're going to extend the interesting ReceiptTypes any time soon, when we do so, it's likely that those two places will need to be extended in the same manner.