fix(chain): sort matches by match_index before groupby

[Toilal]

Reworked, rebased and cleaned-up version of #31 (thanks @denini08 for the catch — credited as co-author).

Problem

itertools.groupby only groups consecutive equal keys. In Chain._group_by_match_index, when matches arrive unordered, groups sharing a match_index get split and the grouped_matches_dict[match_index] = list(match) assignment overwrites the earlier group, silently dropping matches.

Fix

Sort by match_index before grouping — a no-op when the input is already ordered, and consistent with the sorted()-before-groupby() convention already used in rules.py.

sorted_matches = sorted(matches, key=lambda m: m.match_index)
for match_index, match in itertools.groupby(sorted_matches, lambda m: m.match_index):
    ...

What changed vs #31

• Rebased on current develop (the original no longer applied — chain.py/test_chain.py had diverged).
• Wrapped the line (the inline sorted() exceeded the 120-char limit → ruff E501).
• Rewrote the test to pass the gates: typed (-> None), no trailing whitespace, runtime Match import. It fails without the fix (the two match_index=2 matches collapse to one) and passes with it.

Verification

• mypy --strict: clean
• pytest: 188 passed under both the re and regex backends
• ruff / full pre-commit: clean
• Confirmed the new test fails on the pre-fix code

Supersedes #31.

🤖 Generated with Claude Code

[denini08]

@Toilal I'm happy to help!