Fix test_stress_creation_and_deletion

[crusaderky]

• Closes test_stress_creation_and_deletion flaky #5388
• Blocked by and incorporates decide_worker_rootish_queuing_disabled assertion fails when retiring worker #7063

[github-actions]

Unit Test Results

See test report for an extended history of previous test failures. This is useful for diagnosing flaky tests.

       15 files  ±0         15 suites  ±0   6h 26m 55s ⏱️ - 18m 52s
  3 161 tests ±0    3 075 ✔️  - 2    84 💤 +1  2 ❌ +1 
23 388 runs  ±0  22 482 ✔️  - 2  901 💤 +1  5 ❌ +1 

For more details on these failures, see this check.

Results for commit f34b72a. ± Comparison against base commit e0669cd.

♻️ This comment has been updated with latest results.

[crusaderky]

Stress test (19x20 repeats) proves this is very stable now.

[fjetter]

Stress test (19x20 repeats) proves this is very stable now.

Does this mean you ran it 19x20 = 380 times or 19 out of 20 successful runs? 🤔

[crusaderky]

Stress test (19x20 repeats) proves this is very stable now.

Does this mean you ran it 19x20 = 380 times or 19 out of 20 successful runs? thinking

380 times over all CI environments

[crusaderky]

Correction: 2 out of 378 times it failed with

ERROR    asyncio:base_events.py:1707 Task was destroyed but it is pending!
task: <Task pending name='Task-121177' coro=<PeriodicCallback._run() running at /Users/runner/work/distributed/distributed/distributed/compatibility.py:157> cb=[IOLoop.add_future.<locals>.<lambda>() at /Users/runner/miniconda3/envs/dask-distributed/lib/python3.8/site-packages/tornado/ioloop.py:688]>

Both failures are concentrated in the same CI run which is suspicious.
The failure rate is below many other tests - I think we should merge.

[crusaderky]

Merging tomorrow if nobody objects