Use issue_config_time_warning for the unset-loop-scope deprecation

[mathewOracle]

pytest_configure warns via a plain warnings.warn() when asyncio_default_fixture_loop_scope is unset. Per pytest's own docs on Config.issue_config_time_warning, warnings raised this way during pytest_configure can't be captured the way per-test warnings are -- pytest has no way to install a hookwrapper around pytest_configure -- so the warning is silently dropped: it never reaches the warnings summary, and doesn't respect -W/filterwarnings.

$ pytest  # asyncio_default_fixture_loop_scope left unset
2 passed in 0.01s   # no warning shown anywhere, including with -W default

Depending on the caller's global warning filter state it can also raise an uncaught PytestDeprecationWarning straight out of pytest_configure, crashing the run outright (reproducible via pytester's in-process runpytest(), which shares the outer process's filters).

Config.issue_config_time_warning is pytest's documented mechanism for exactly this case, and is what pytest's own core plugins use for their config-time deprecations (e.g. _pytest/pastebin.py). Switching to it makes the warning show up in the warnings summary and respect the caller's filters, with no change to the message or when it fires.

Verified live against the exact repro from the issue: the warning now appears in a plain pytest run with no flags needed. Added a regression test that fails against the previous code (either silently, or via INTERNALERROR depending on inherited filters) and passes with this change, plus a sanity test that no warning fires when the option is configured. Full test suite shows identical pre-existing failures/errors before and after this change (unrelated compatibility gaps with the installed pytest version); no new failures introduced. ruff check clean.

Closes #1142

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 95.29%. Comparing base (dcd61c2) to head (6014a08).
⚠️ Report is 6 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1557      +/-   ##
==========================================
+ Coverage   94.50%   95.29%   +0.78%     
==========================================
  Files           2        2              
  Lines         510      510              
  Branches       62       62              
==========================================
+ Hits          482      486       +4     
+ Misses         22       20       -2     
+ Partials        6        4       -2     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[tjkuson]

I think we can delete this comment. It reads more like developer commentary pertaining to this PR rather than anything useful long-term.

[tjkuson]

Can we simplify these changes? Some of the assertions pass on main and, as-written, obfuscates the behaviour we're trying to protect. I'd also expect a test that covers suppressing this warning through -W and filterwarnings.

[tjkuson]

Can we make this fragment more precise (a user could mistakenly interpret this as thinking we haven't been emitting this warning at all) and remove the implementation detail about Config.issue_config_time_warning?

[mathewOracle]

Thanks for the review @tjkuson! Pushed 6014a08 addressing all three points:

• Comment removed — agreed, it was PR commentary rather than durable documentation. The rationale now lives in the commit message and the issue.
• Tests simplified — dropped the session-scoped fixture/loop machinery that wasn't relevant. There are now two focused tests: one asserting the warning reaches the warnings summary, and a parametrised one covering suppression via both -W on the command line and filterwarnings in the ini. I checked each assertion fails against main (all three fail with INTERNALERROR ... PytestDeprecationWarning), so nothing passes for free.
• Changelog reworded — drops the Config.issue_config_time_warning implementation detail and describes the user-visible behaviour instead, without implying the warning was never emitted.

Note: tests/test_set_event_loop.py::test_asyncio_run_after_async_fixture_does_not_leak_loop fails on my machine, but it also fails on main (dcd61c2) — unrelated to this change.