feat(event cache): redact previously-seen events in the linked chunk too

[bnjbvr]

This makes it possible to:

• replace one item in a linked chunk, and have all the consumers handle the update. This will be useful for the second part of the PR (applying redactions to the linked chunk), and in the future to replace decrypted events (when decryption happens in the event cache).
• redact, in memory and in storage, an event by its redacted form, so that it's not possible to read it on disk.

We could add a migration to apply redactions for existing users of the nightly cache. I'd rather do this as a followup, since this PR is growing a bit large already (and another possibility is also to clear all events stored in the event cache, as a dirty yet safe migration — we might need this because I spotted an opportunity to not persist unneeded data from storage).

[codecov]

Codecov Report

Attention: Patch coverage is 74.77477% with 28 lines in your changes missing coverage. Please review.

Project coverage is 85.39%. Comparing base (c969f90) to head (17db2fb).
Report is 20 commits behind head on main.

Files with missing lines	Patch %	Lines
...tes/matrix-sdk-ui/src/timeline/controller/state.rs	37.50%	10 Missing ⚠️
crates/matrix-sdk-common/src/linked_chunk/mod.rs	78.94%	4 Missing ⚠️
...s/matrix-sdk-common/src/linked_chunk/relational.rs	50.00%	4 Missing ⚠️
crates/matrix-sdk/src/event_cache/room/events.rs	86.95%	3 Missing ⚠️
crates/matrix-sdk/src/event_cache/room/mod.rs	72.72%	3 Missing ⚠️
crates/matrix-sdk-ui/src/timeline/event_handler.rs	50.00%	2 Missing ⚠️
crates/matrix-sdk-base/src/rooms/normal.rs	50.00%	1 Missing ⚠️
...es/matrix-sdk-common/src/deserialized_responses.rs	80.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4536      +/-   ##
==========================================
- Coverage   85.43%   85.39%   -0.05%     
==========================================
  Files         285      285              
  Lines       32112    32205      +93     
==========================================
+ Hits        27435    27501      +66     
- Misses       4677     4704      +27     

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

[Hywan]

👍

[Hywan]

Push the update after current_items[item_index] = item;: do the action before triggering the update.

[bnjbvr]

This means we have to clone before, and the original thing is unused if self.updates.as_mut().is_none(). Plus, the updates aren't reflected in real-time, they're buffered until they're consumed, as far as I can tell? So this should be fine to keep as is. I can add a comment explaining that this avoids a clone if you'd like.

[andybalaam]

Looks good. I'm not hugely confident I understand all the implications, but the code looks good. Some small suggestions and questions, nothing blocking.