fix(pinned events): make sure to initialize ordering before reloading events from disk

[bnjbvr]

See the commit description. Honestly, even after writing it myself, I do find that API is a bit of a mess; relying on the linked chunk updates emitted by the same linked chunk might be too hacky for the event ordering to function properly overall. But first, let's fix the bug.

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

[codecov]

Codecov Report

❌ Patch coverage is 54.32099% with 37 lines in your changes missing coverage. Please review.
✅ Project coverage is 89.89%. Comparing base (8b14a89) to head (4892c81).
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
crates/matrix-sdk/src/event_cache/persistence.rs	53.84%	32 Missing and 4 partials ⚠️
...ix-sdk/src/event_cache/caches/pinned_events/mod.rs	66.66%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #6466      +/-   ##
==========================================
+ Coverage   89.88%   89.89%   +0.01%     
==========================================
  Files         380      380              
  Lines      105023   105026       +3     
  Branches   105023   105026       +3     
==========================================
+ Hits        94395    94416      +21     
+ Misses       7009     6993      -16     
+ Partials     3619     3617       -2     

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

[codspeed-hq]

Merging this PR will not alter performance

✅ 50 untouched benchmarks

---

_{Comparing bnjbvr/pinned-events-ordering-tracker (4892c81) with main (8b14a89)}

[Hywan]

Do we have a failing test to ensure this patch fixes it?

[bnjbvr]

The answer is in the second commit's message:

This fixes a subtle bug, found by re-enabling the assertion disabled in github.com/#5332.
[...]
it's hard to make a standalone test case for this, so I'll really on people re-enabling the ordering assertion back in the future

[bnjbvr]

Fwiw, I've tried to write a failing test for this since the previous message, but I've realized that the ordering might not even be used in the case of the pinned events, so this is only fixing the assertion re-enabled in the test anyways. In other words: there's very little chance that this affects real-world users, at this point.