feat(event cache): process room updates concurrently

[bnjbvr]

This should improve latency of event cache updates in general, especially for massive initial sync responses (which may include many many rooms).

[codecov]

Codecov Report

Attention: Patch coverage is 50.00000% with 6 lines in your changes missing coverage. Please review.

Project coverage is 88.87%. Comparing base (165ec9d) to head (3abf5a9).

✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
crates/matrix-sdk/src/event_cache/mod.rs	50.00%	2 Missing and 4 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #5426      +/-   ##
==========================================
+ Coverage   88.85%   88.87%   +0.02%     
==========================================
  Files         333      333              
  Lines       91244    91256      +12     
  Branches    91244    91256      +12     
==========================================
+ Hits        81072    81103      +31     
+ Misses       6348     6326      -22     
- Partials     3824     3827       +3     

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

[poljar]

Performance improvements without numbers make me raise an eyebrow, but ok it's one of the simpler changes.

[bnjbvr]

Annnnnnnnnnnnnnnnnnnd it's not a speedup, for reasons I can't fathom (cross-process lock contention? some other lock contention? benchmark doesn't do what we think it does?), and I don't have much time to investigate unfortunately, having to pivot to more oriented feature work.

For posterity, the benchmark added in #5435 shows the following perf diff, with this patch:

reading/handle_room_updates/sqlite store/1
                        time:   [2.2372 ms 2.4769 ms 2.6958 ms]
                        thrpt:  [370.95  elem/s 403.73  elem/s 446.98  elem/s]
                 change:
                        time:   [-20.865% -13.429% -4.3047%] (p = 0.01 < 0.05)
                        thrpt:  [+4.4983% +15.512% +26.366%]
                        Performance has improved.
Found 1 outliers among 10 measurements (10.00%)
  1 (10.00%) high mild
reading/handle_room_updates/sqlite store/10
                        time:   [44.909 ms 46.289 ms 47.953 ms]
                        thrpt:  [208.54  elem/s 216.03  elem/s 222.67  elem/s]
                 change:
                        time:   [-9.6624% +0.9193% +15.623%] (p = 0.91 > 0.05)
                        thrpt:  [-13.512% -0.9109% +10.696%]
                        No change in performance detected.
Found 2 outliers among 10 measurements (20.00%)
  1 (10.00%) low mild
  1 (10.00%) high severe
Benchmarking reading/handle_room_updates/sqlite store/100: Warming up for 3.0000 s
Warning: Unable to complete 10 samples in 5.0s. You may wish to increase target time to 5.2s.
reading/handle_room_updates/sqlite store/100
                        time:   [484.03 ms 496.08 ms 505.25 ms]
                        thrpt:  [197.92  elem/s 201.58  elem/s 206.60  elem/s]
                 change:
                        time:   [+56.456% +65.016% +74.717%] (p = 0.00 < 0.05)
                        thrpt:  [-42.765% -39.400% -36.084%]
                        Performance has regressed.
Found 1 outliers among 10 measurements (10.00%)
  1 (10.00%) low mild
Benchmarking reading/handle_room_updates/sqlite store/1000: Warming up for 3.0000 s
Warning: Unable to complete 10 samples in 5.0s. You may wish to increase target time to 69.8s.
reading/handle_room_updates/sqlite store/1000
                        time:   [4.1170 s 4.1930 s 4.2634 s]
                        thrpt:  [234.56  elem/s 238.49  elem/s 242.90  elem/s]
                 change:
                        time:   [+41.880% +45.468% +49.428%] (p = 0.00 < 0.05)
                        thrpt:  [-33.078% -31.256% -29.518%]
                        Performance has regressed.
Benchmarking reading/handle_room_updates/sqlite store/10000: Warming up for 3.0000 s^C⏎