Add back synced frame method and use it for triggerRepaint.

[xabbu42]

Launch Checklist

An experimental fix for #4534. With this I can again a capture a single frame with spector.js, as all the rendering is directly done inside the requestAnimationFrame callback.

Unfortunatly tests dont run through but fail after some random point. Probably a race I introduced?

• Confirm your changes do not include backports from Mapbox projects (unless with compliant license) - if you are not sure about this, please ask!
• Briefly describe the changes in this PR.
• Link to related issues.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 65.29%. Comparing base (ba721b4) to head (ca568bf).
Report is 7 commits behind head on main.

❗ There is a different number of reports uploaded between BASE (ba721b4) and HEAD (ca568bf). Click for more details.

HEAD has 1 upload less than BASE

Flag	BASE (ba721b4)	HEAD (ca568bf)
	9	8

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #4535       +/-   ##
===========================================
- Coverage   87.79%   65.29%   -22.50%     
===========================================
  Files         246      246               
  Lines       33450    33457        +7     
  Branches     2219     1455      -764     
===========================================
- Hits        29367    21847     -7520     
- Misses       3081    10765     +7684     
+ Partials     1002      845      -157     

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

[HarelM]

requestAnimationFrame is not a method that runs right away as far as I know, it might have higher priority than a promise, but I'm not sure I saw it anywhere, so I'm not sure I know how this fix will actually solve anything, but let me know if it does.
I think the code introduced here is fine in terms of how it looks.
A test is needed to cover the relevant issue (spector.js if I understood correctly).

In case this is not ready for review, please change it to a draft and ping me when you change it back to regular PR.

THANKS!

[xabbu42]

this experiment is not about the time between the call to requestAnimationFrame and when the passed callback is called, but about the time between the call to the passed callback and the actual map._render call. Currently the passed callback only schedules the _render and then returns whereas before the callback directly called _render.

Come to think of it, it should be possible to keep with the promise based interface for frameAsync for the time between the call to frameAsync and when the callback of requestAnimationFrame actually gets triggered while at the same time calling _render directly in the callback. I'll try to implement that approach.

[HarelM]

Can we pass down the callback to the request animation frame and then call the resolve at the end? Would that work? Or is this the scratched out part in your comment?