Wrap product() in list() where passed to parametrize

[Sanjays2402]

What

Two tests pass an itertools.product object directly to
@pytest.mark.parametrize:

• tests/test_bidict.py::test_eq_and_or_with_non_mapping
• tests/test_bidict.py::test_putall_matches_bulk_put

product(...) returns a one-shot iterator, not a Collection. This
change wraps both in list() so the argvalues are a concrete
Collection.

Why

Starting with pytest 9.1, passing a non-Collection iterable to
parametrize raises PytestRemovedIn10Warning, and pytest 10 removes
support entirely — at which point collecting tests/test_bidict.py
will fail. See the pytest deprecation note:
https://docs.pytest.org/en/stable/deprecations.html#parametrize-iterators

Today (pytest 8.x/9.x) the suite still collects, but the run already
emits the warning:

PytestRemovedIn10Warning: Passing a non-Collection iterable to parametrize is deprecated.
  Test: tests/test_bidict.py::test_eq_and_or_with_non_mapping, argvalues type: product
  Test: tests/test_bidict.py::test_putall_matches_bulk_put, argvalues type: product

Reproduce (deterministic)

Promote the warning to an error exactly as pytest 10 will:

$ python -m pytest tests/ -W "error::pytest.PytestRemovedIn10Warning"
E   pytest.PytestRemovedIn10Warning: Passing a non-Collection iterable to parametrize is deprecated.
E   Test: tests/test_bidict.py::test_eq_and_or_with_non_mapping, argvalues type: product
ERROR tests/test_bidict.py - pytest.PytestRemovedIn10Warning: ...
!!!!! Interrupted: 1 error during collection !!!!!
1 error in 0.09s

Before / after

	before (product(...))	after (list(product(...)))
pytest -W error::pytest.PytestRemovedIn10Warning	ERROR at collection	114 passed
default pytest run	2 PytestRemovedIn10Warning	0 warnings

Notes

• Behavior of the parametrization is unchanged today; this only
  materializes the argvalues into a concrete Collection.
• Mirrors the existing list(powerset(product(ks, vs))) idiom already
  used in this module.
• No CHANGELOG.rst entry: this is a test-tooling-only forward-compat
  fix with no user-facing change, consistent with recent bug-fix PRs
  (Fix TypeError when comparing OrderedBidict keys() with items() or vice versa #376, Fix _dedup no-op detection for non-reflexive-equality keys/values (e.g. NaN) #377) which touched only source + tests.
• Full suite green with the project's own config: 114 passed.
  ruff check clean on the changed file.

[github-actions]

Benchmark report

✅ No threshold-triggering regressions detected.

• Baseline asset: GHA-linux-cachegrind-x86_64-CPython-3.14-baseline.json
• Threshold: min:5%
• Result: no threshold-triggering regressions detected
• Run: benchmark workflow run

This benchmark check is informational and does not block merging.

[jab]

Thanks for the contribution!