fix(ffi): propagate initial values before the future is picked by a runtime

[bnjbvr]

Same spirit as #4818; doing work after the async move "moves" the code inside the future, and it will be executed only when a tokio runtime is available, instead of doing it right away. We can improve and do it right away in a few cases:

• when getting the initial timeline items
• when looking at the room directory search results

I've audited all the spawns at the FFI layer, so we should be in a good shape now.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 86.46%. Comparing base (5acaaf5) to head (791e0f8).
Report is 60 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #4820   +/-   ##
=======================================
  Coverage   86.45%   86.46%           
=======================================
  Files         297      297           
  Lines       34565    34565           
=======================================
+ Hits        29884    29885    +1     
+ Misses       4681     4680    -1     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Hywan]

Yup, makes sense! Thanks for the fixes.

[Hywan]

I honestly don't understand the error on Complement Crypto.

https://github.com/matrix-org/complement-crypto/blob/23821cf6320beb15b20233dc5728a019fcc02f15/internal/api/rust/rust.go#L720-L721

The listener should be called the same way. If we are hitting a timeout, it means the listener isn't called, which is surprising as it should be called… faster.

[bnjbvr]

Kegan reported that the complement-crypto test expects to get the first events after AddTimelineListener has returned, not while it happens, so it's likely deadlocking somewhere. @kegsay would you kindly help us and rework the Go code so it works with this breaking change, please? Otherwise we can also disable the complement-crypto testing in the meanwhile.

[Hywan]

Re-run the complement crypto job now there is matrix-org/complement-crypto#169.