Long lived branch with Overlay refinements

[rjrjr]

Holding off on merging into main until I have more confidence in the changes, which I'm getting by applying them across Square's flagship Android apps; and until I'm ready to deal with the recent coroutines version update (#817).

Introduces OverlayDialogHolder.onUpdateBounds

OverlayDialogHolder.onUpdateBounds replaces ScreenOverlayDialogFactory.updateBounds.

I finally tried to use ScreenOverlayDialogFactory.updateBounds and was reminded once again why it's bad to create interfaces that build something and then take it back later to be updated. If you want the renderings to be involved in setting the bounds policy, there was no way to smuggle that information along with the Dialog returned from buildDialogWithContent, so that updateBounds could honor it. Also, it was always pretty ugly that the bounds hook was special to the ScreenOverlay world.

The fix is make bounds maintenance part of the job of the OverlayDialogHolder interface. Square reviewers will notice that we're now an even more faithful rip off of Mosaic's DialogRunner -- hi @helios175.

No more hard coded EnvironmentScreen in WorkflowLayout.

WorkflowLayout was wrapping things in EnvironmentScreen for no real reason. No longer.

[rjrjr]

@steve-the-edwards Uh oh:

com.squareup.benchmarks.performance.complex.poetry.RenderPassTest > renderPassCounterFrameTimeoutComplexNoInitializingStateHighFrequencyEvents[test(AVD) - 10] FAILED 
	java.lang.AssertionError: Uh Oh! You have worsened the the number of Render Passes (lower better) by 2 (2.06%) for Runtime: FrameTimeout; the 'Raven navigation (no initializing state) scenario with high frequency events'! The value is now 99 (was 88..97). This likely results in worse performance. You can check with the timing benchmarks if you disagree.
	at org.junit.Assert.fail(Assert.java:89)

I'm surprised by this, and I don't see what I've done to cause it. Can I just bump the expected range, or do we need to dig in?

[rjrjr]

Re-running that shard to see if this is flaky.

[rjrjr]

Sure enough, it was a flake. :(

[rjrjr]

This is ready for review, although I might not merge it until I get to exercise it a bit harder as a snapshot (1.8.0-rjrjr2beta06-SNAPSHOT).

[steve-the-edwards]

Sure enough, it was a flake. :(

If that was just a flake/within reasonable bounds I am fine with bumping up the expected range.

[vgonda]

:stamp:

[rjrjr]

I'm going to merge #840 into this branch, and hold off on merging into main until I have a bit more confidence in this series of changes. I'm already working on a refinement of our back button handling in Overlay, and expect to float another PR targeting this branch today.

[rjrjr]

I'm going to merge #840 into this branch, and hold off on merging into main until I have a bit more confidence in this series of changes. I'm already working on a refinement of our back button handling in Overlay, and expect to float another PR targeting this branch today.

#840 is being weird in CI, though everything passes locally and all our internal tests like it. I'm going to go ahead and merge this branch and target #840 at main, hoping that makes CI sweet again.