The upgrade wait test gives PowerShell 180 seconds to exit, the same limit the deploy tests use

[erikdarlingdata]

What failed

On #4896's CI run (commit 950bb7db, Darling PG tests (1) on Windows), DarlingUpgradeInstallTreeWaitTests.WaitDarlingInstallTreeClear_ReturnsEmptyAfterTheProcessExits_ButReturnsItWhenTheBoundIsTooShort failed with pwsh did not exit within 60s running the extracted wait function. That shard ran 595 s, longer than usual, so the runner was loaded. #4896 doesn't touch the upgrade script or this test.

Why

The test runs the extracted Wait-DarlingInstallTreeClear function in Windows PowerShell 5.1 on Windows, with bounds of 10 s and 1 s. Its own limit for the PowerShell process to exit was 60 s. On a loaded Windows runner, PowerShell 5.1's start and module load alone can take most of that. The deploy tests hit the same thing, and #4866 raised their limit to 180 s (PowerShellExitLimit).

What changed (tests only)

• The limit: the test's exit limit for the PowerShell process is 180 s (PowerShellExitLimit), the same as the deploy tests.
• On a timeout: the process tree is killed, as the deploy tests do.
• Nothing else changed: the bounds the test drives (10 s and 1 s) and every assertion on the function's result are the same, so a wait that doesn't clear, or one that waits past its bound, still fails.

Testing

• Darling.Tests builds with 0 warnings and 0 errors.
• CI on this branch runs the test on the Windows PG shards.

CHANGELOG

None: a test-only change.