Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com>

* test(parse): add assertion for v6 inline reference behavior

Verify that inline references are captured in the references array,
documenting the behavioral change from v5 where they remained in body.

* test(parse): document complete v6 inline reference behavior

Add assertions to verify that:
- Both reference instances are captured (2 total)
- Body content with references moves to footer
- Each reference object includes the raw source line

This comprehensively documents the behavioral change from v5
(references stayed in body) to v6 (references move to footer).

* test(parse): split inline reference test into two focused tests

Separate concerns for better clarity:
- Test 1: Inline reference parsing (verifies v6 moves body refs to footer)
- Test 2: Comment line filtering (verifies commentChar option works)

This addresses co-maintainer feedback about the confusing mixed test.

* Revert "refactor(ensure): replace all lodash string methods with kasi and manual (#4602)"

This reverts commit a1ef07e.

[email protected] has a bug where it incorrectly identifies lowercase Cyrillic
text as PascalCase/TitleCase, causing the subject-case rule to reject
valid lowercase subjects in non-Latin scripts.

Restores lodash-based case detection which correctly handles non-Latin
alphabets including Cyrillic, Chinese, Arabic, and Hebrew.

Related to #4620

* test(rules): add non-Latin script coverage for subject-case rule

Add regression tests for lowercase and uppercase subjects in:
- Cyrillic (Russian)
- Chinese
- Arabic
- Hebrew
- Mixed Latin + Cyrillic

These tests ensure the subject-case rule correctly handles non-Latin
alphabets and prevent future regressions like the [email protected] bug.

Fixes #4620