Make xfail strict

[Pierre-Sassoulas]

Type of Changes

	Type
✓	🔨 Refactor

Description

Sets xfail_strict = true for pytest.

Right now an xfail that starts passing is silently absorbed: pytest reports
XPASS, the suite stays green, and the mark stays in the file. So the test goes
on being advertised as expected-to-fail long after it works, and the next person
to read it has no reason to doubt that. With xfail_strict an unexpected pass
fails the suite, which is the only moment anyone would think to delete the mark.

This came up on #3211, where two xfails were added for known holes. Non-strict,
they would have outlived the fixes.

No test needed changing. All 15 xfails still fail, on 3.13 and on 3.14:

2080 passed, 61 skipped, 15 xfailed     # 3.13
2050 passed, 90 skipped, 15 xfailed     # 3.14

I can only check Linux locally, so CI is the real test for the two marks whose
outcome could plausibly differ elsewhere --- c_buffer in
tests/brain/test_ctypes.py and test_factory_methods_inside_binary_operation.
If either turns up as XPASS on Windows or PyPy the fix is a condition on the
mark, not a revert.

Noticed while here, not touched: the reason on
test_factory_methods_inside_binary_operation still says "cannot be inferred on
Python 3.8", which astroid has not supported for a while. The test does still
fail, so the mark is right and only the reason is stale.

[codspeed-hq]

Merging this PR will not alter performance

✅ 3 untouched benchmarks
⏩ 1 skipped benchmark¹

---

_{Comparing xfail-strict (f3e3014) with main (62ef6b3)}

Footnotes

1. 1 benchmark was skipped, so the baseline result was used instead. If it was deleted from the codebase, click here and archive it to remove it from the performance reports. ↩

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.65%. Comparing base (62ef6b3) to head (f3e3014).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #3232   +/-   ##
=======================================
  Coverage   93.65%   93.65%           
=======================================
  Files          93       93           
  Lines       11583    11583           
=======================================
  Hits        10848    10848           
  Misses        735      735           

Flag	Coverage Δ
linux	93.51% <ø> (ø)
pypy	93.65% <ø> (ø)
windows	93.62% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

🚀 New features to boost your workflow:

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