Fix flaky HTTP/2 test

[halter73]

I'm taking my own advice from #39588 (comment) and ensuring we lose the AsyncTestSyncContext for the duration of the Kestrel's in-memory HTTP/2 tests. Previously we were just doing this for the duration of InitializeConnectionAsync which fixed some flakiness.

By ensuring the echo app in the AcceptNewStreamsDuringClosingConnection test runs inline with await SendDataAsync(3, _helloBytes, true); and completes, we can ensure the final goaway gets sent without resorting to hacks like calling TriggerTick() in a loop until a timeout.

I'm not sure how applicable this fix will be after @davidfowl's #40925. I think it should still be good because Http2FrameWriter's channel runs inline in tests.

Fixes (but not really fixes because unquarantine rules) #39479 #41172

[BrennanConroy]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 3 pipeline(s).

[BrennanConroy]

This PR seems to have made stuff flaky? Or for some reason flaky tests only show up with this change 😆

The three CI runs have all had different kestrel tests fail.
https://dev.azure.com/dnceng/public/_build/results?buildId=1717072&view=ms.vss-test-web.build-test-results-tab
https://dev.azure.com/dnceng/public/_build/results?buildId=1716614&view=ms.vss-test-web.build-test-results-tab
https://dev.azure.com/dnceng/public/_build/results?buildId=1742997&view=ms.vss-test-web.build-test-results-tab

[ghost]

Looks like this PR hasn't been active for some time and the codebase could have been changed in the meantime.
To make sure no breaking changes are introduced, please remove the pending ci rerun label to kick off a new CI run.

[amcasey]

Triage: Closing this for bookkeeping reasons. Feel free to re-open if we find a way to reduce the flakiness.

[ghost]

Hi @amcasey. It looks like you just commented on a closed PR. The team will most probably miss it. If you'd like to bring something important up to their attention, consider filing a new issue and add enough details to build context.