Remove is_python_shutting_down

[hendrikmakait]

This PR binds sys.is_finalizing to make sure that it's always available during interpreter shutdown and removes the (to the best of my knowledge superfluous) is_python_shutting_down.

This handles errors during shutdown like

Exception ignored in: <function Future.__del__ at 0x6f23b140ccc0>
Traceback (most recent call last):
  File "/usr/local/python/python3/std/lib64/python3.11/site-packages/distributed/client.py", line 510, in __del__
TypeError: 'NoneType' object is not callable

• 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 18m 15s ⏱️ + 3m 1s
 4 080 tests ±0   3 958 ✅ ±0    112 💤 ±0   9 ❌ ±0  1 🔥 ±0 
55 188 runs   - 1  52 741 ✅  - 2  2 435 💤 ±0  11 ❌ +1  1 🔥 ±0 

For more details on these failures and errors, see this check.

Results for commit b9b1678. ± Comparison against base commit 50700f3.

♻️ This comment has been updated with latest results.

[hendrikmakait]

cc @rjzamora in case RAPIDS uses is_python_shutting_down somewhere internally.

[hendrikmakait]

This isn't great, but it's the safest way of checking whether the interpreter is shutting down. More concretely, whether it's safe to use threads. Threads are being shutdown before atexit hooks so we need a mechanism like is_python_shutting_down that becomes the first atexit hook in distributed and lets us know whether we have entered shutdown, or we need to propagate that knowledge during shutdown, or we need to check whether it's safe to use threads where necessary (as done here).

[fjetter]

what a mess. I'm fine with this