Ensure that adaptive only stops once

[hendrikmakait]

On main, Adaptive could get stopped multiple times, issuing an Adaptive stop log message each time. This could get confusing if adaptive is replaced because the stop logs of the old Adaptive might appear after the start log of the new one.

Example

2024-07-31 09:13:47,960 - distributed.deploy.adaptive_core - INFO - Adaptive stop
2024-07-31 09:13:47,960 - distributed.deploy.adaptive - INFO - Adaptive scaling started: minimum=5 maximum=990
2024-07-31 09:13:48,460 - distributed.deploy.adaptive_core - INFO - Adaptive stop

This PR avoids that.

• Tests added / passed
• Passes pre-commit run --all-files

[github-actions]

Unit Test Results

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

    29 files  ± 0      29 suites  ±0   11h 52m 59s ⏱️ - 2m 1s
 4 093 tests + 1   3 981 ✅ +10    111 💤 ±0  1 ❌  - 9 
55 367 runs  +14  52 936 ✅ +22  2 430 💤 +1  1 ❌  - 9 

For more details on these failures, see this check.

Results for commit 6c01e8f. ± Comparison against base commit c44ad22.

♻️ This comment has been updated with latest results.

[jrbourbeau]

FWIW it looks like the windows failures are related to the changes here

Traceback (most recent call last):

  File "C:\Users\runneradmin\miniconda3\envs\dask-distributed\Lib\site-packages\tornado\ioloop.py", line 750, in _run_callback

    ret = callback()

          ^^^^^^^^^^

  File "D:\a\distributed\distributed\distributed\deploy\adaptive_core.py", line 151, in _start

    if self._state != "starting":

       ^^^^^^^^^^^

AttributeError: 'Adaptive' object has no attribute '_state'

[hendrikmakait]

Good catch, thanks @jrbourbeau! This should now be fixed. I'm a little stunned by the fact that this didn't show up in the test summary.