ui: Redact replied-to event inside in_reply_to in timeline

[jplatte]

… and more in-reply-to / redaction improvements.

Ready for testing in EX, needs one or two tests before merging.

[codecov]

Codecov Report

Patch coverage: 53.21% and project coverage change: -0.51% ⚠️

Comparison is base (0a361ef) 77.26% compared to head (1c8017a) 76.75%.

❗ Current head 1c8017a differs from pull request most recent head d8631a4. Consider uploading reports for the commit d8631a4 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2307      +/-   ##
==========================================
- Coverage   77.26%   76.75%   -0.51%     
==========================================
  Files         171      171              
  Lines       18207    18237      +30     
==========================================
- Hits        14067    13998      -69     
- Misses       4140     4239      +99     

Files Changed	Coverage Δ
...s/matrix-sdk-ui/src/timeline/event_item/content.rs	54.12% <30.30%> (-7.49%)	⬇️
crates/matrix-sdk-ui/src/timeline/inner.rs	74.59% <69.23%> (+1.34%)	⬆️
...rates/matrix-sdk-ui/src/timeline/event_item/mod.rs	73.64% <88.88%> (+1.14%)	⬆️
crates/matrix-sdk-ui/src/timeline/event_handler.rs	78.14% <100.00%> (+0.50%)	⬆️
crates/matrix-sdk-ui/src/timeline/traits.rs	53.19% <100.00%> (+4.35%)	⬆️

... and 9 files with indirect coverage changes

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

[jplatte]

Tests have been written, this is ready for review now.

[bnjbvr]

Could we store some sort of id or index that uniquely and permanently references the actual item, that would allow to not duplicate the item in the timeline? Then we get rid of a whole range of state duplication bugs (like the one fixed by the first commit of that PR, IIUC).

[jplatte]

It sounds like you want more fine-grained reactivity? As long as the API consumer expects ObservableVector<Arc<TimelineItem>> level updates for every affected timeline item, we'll fundamentally have to walk the list to find out which items need to be updated. Let's discuss this on Matrix.

[bnjbvr]

We chatted about this, and there might be a solution where we (at least internally) have some reactive mapping of timeline internal id -> timeline item. Without changing the external API, this would allow to avoid iterating over all the timeline items to propagate edits/redactions. But this would require something like an ObservableIndexMap or similar, which doesn't exist yet. So let's take this PR in the meanwhile.