perf(search): cache the tantivy IndexReader on RoomIndex

[procr1337]

This is a fix for a performance issue we encountered when backfilling events into the search engine in a real matrix room with 10k+ messages. The issue is that RoomIndex::bulk_execute performs reading operations between each write (e.g. inside ::add(), it calls ::contains(), which transitively calls ::get_reader()). The reader creation would take about ~30ms in this case per inserted event, which is highly prohibitive.

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

[codecov]

Codecov Report

❌ Patch coverage is 90.90909% with 1 line in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (main@023f5bd). Learn more about missing BASE report.
⚠️ Report is 6 commits behind head on main.
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
crates/matrix-sdk-search/src/index/mod.rs	90.90%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #6707   +/-   ##
=======================================
  Coverage        ?   89.95%           
=======================================
  Files           ?      397           
  Lines           ?   110660           
  Branches        ?   110660           
=======================================
  Hits            ?    99539           
  Misses          ?     7351           
  Partials        ?     3770           

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

[procr1337]

@stefanceriu the codecov report seems broken, it doesn't show any info about covered or uncovered lines. It just says "No Files covered by tests were changed"

[codspeed-hq]

Merging this PR will not alter performance

✅ 50 untouched benchmarks

---

_{Comparing procr1337:search-reader-recreation-fix (0346d34) with main (023f5bd)}

[stefanceriu]

Looks great to me, thank you for taking the time!