Correctly reset log level when temporarily changing it

[MetRonnie]

Fixes tests that were failing on my cylc remove branch due to a change in order

Context

Running these 2 tests in this order would guarantee test_Timer would fail.

pytest -n 0 \
  tests/integration/scripts/test_show.py::test_workflow_meta_query \
  tests/unit/test_timer.py::test_Timer

This is due to the way we were temporarily changing the Cylc logger level during Scheduler.start()

cylc-flow/cylc/flow/loggingutil.py

Lines 426 to 437 in cadf1ef

	@contextmanager
	def patch_log_level(logger: logging.Logger, level: int = logging.INFO):
	"""Temporarily patch the logging level of a logger if the specified level
	is less severe than the current level.
	Defaults to INFO.
	"""
	orig_level = logger.getEffectiveLevel()
	if level < orig_level:
	logger.setLevel(level)
	yield
	logger.setLevel(orig_level)

Logger.getEffectiveLevel() will return the root logger level if Logger.level is NOTSET.

In pytest the root logger level is WARNING, so in the tests, there were cases where the Cylc logger was being incorrectly reset to WARNING instead of NOTSET

Check List

• I have read CONTRIBUTING.md and added my name as a Code Contributor.
• Contains logically grouped changes (else tidy your branch by rebase).
• Does not contain off-topic changes (use other PRs for other changes).
• No dependency changes
• Tests are included
• No changelog entry needed as not affecting users
• No docs needed
• If this is a bug fix, PR should be raised against the relevant ?.?.x branch.

[MetRonnie]

Pointless comparison of <function logger.getEffectiveLevel> != <int>

[wxtim]

Looks good - one question about whether both tests are necessary should not stand in the way of merger.

[oliver-sanders]

merger-ing