[Bug]: stuck executions with parallel/map operation

[zhongkechen]

Expected Behavior

The executions with map and parallel always make progresses when branches complete

Actual Behavior

The test cases from this PR sometimes get stuck:
#284

Steps to Reproduce

[SilanHe edits]
If you run this many times in your IDE, you can easily reproduce this for the "stuck" case. I'm using IntelliJ and breakpointing the SDK code where it's "registering" and "deregistering" the code. In the slower debugger environment, the code breaks down and often is stuck with child context threads unable to proceed.

SDK Version

1.0.0

Java Version

17

Is this a regression?

No

Last Working Version

No response

Additional Context

No response

[SilanHe]

Steps to Reproduce

[SilanHe edits]
If you run this many times in your IDE, you can easily reproduce this for the "stuck" case. I'm using IntelliJ and putting breakpoints at areas in the code where we are managing the threads. That is, whenever we are "registering" or "deregistering" a thread, I am putting a breakpoint. Stepping through the code, you can easily reproduce the stuck execution as the changes in timing causes the execution to be stuck.

[Update] There's another issue where the test can sometimes return "Pending" executionStatus.

I'm using IntelliJ and editing the run configuration for the testParallelWithMinSuccessful_earlyTermination test to "run until failure".

[SilanHe]

The code is "breaking" here because while the ExecutionStatus is in "Pending", there are no waits or retries to advance so advanceReadyOperations returns false.