Fix crash when opening two platform view pages one after another in hybrid composition mode

[0xZOne]

Try to fix issue: #89408

Consider sharing engine scenario, when the previous FlutterView detaches from the engine, the overlay Surface cached in the PlatformViewController destroyed, and the corresponding surface in SurfacePool is still alive and used for the current FlutterView. This inconsistency causes an assertion crash.

To make SurfacePool work for the shared engine scenario, we completely here hand over the responsibility of destroying the overlay surface cached by PlatformViewsController to SurfacePool.

Pre-launch Checklist

• I read the Contributor Guide and followed the process outlined there for submitting PRs.
• I read the Tree Hygiene wiki page, which explains my responsibilities.
• I read and followed the Flutter Style Guide and the C++, Objective-C, Java style guides.
• I listed at least one issue that this PR fixes in the description above.
• I added new tests to check the change I am making or feature I am adding, or Hixie said the PR is test-exempt. See testing the engine for instructions on
  writing and running engine tests.
• I signed the CLA.
• All existing and new tests are passing.

If you need help, consider asking for advice on the #hackers-new channel on Discord.

[chinmaygarde]

cc @blasten

[blasten]

This seems more of a general issue. shouldn't each engine have its own SurfacePool? @gaaclarke

[gaaclarke]

I don't see a reason why lightweight engines would need their own SurfacePool. @xster implemented this and systematically looked at all the resources resident in memory and shared them when possible. Lightweight engines share threads so they will be communicating with this pool serially, so there is no need to have multiple pools.

[0xZOne]

This seems more of a general issue. shouldn't each engine have its own SurfacePool? @gaaclarke

Here is the case where multiple Flutter Views reuse a single engine.

[chinmaygarde]

@blasten & @gaaclarke are you able to suggest a resolution to @0xZOne s issue without needing to creating separate pools? @xster is probably no longer owning these components. So we have to decide how to proceed with issues raised here.

[zanderso]

@blasten @gaaclarke From PR triage, it looks like the status of this PR is an open question for you from @chinmaygarde . PTAL, thanks!

[chinmaygarde]

Any guidance here? If not, a short term solution to fix the crash would be to go with the separate pools.

[0xZOne]

Fixed by #28894