Exclude separating lines from row index validation

[kokotatan]

Fixes #464

A sized custom showindex rejects valid input when SEPARATING_LINE rows are present because its length is compared with all display rows. A DataFrame's implicit index also shifts separator labels onto subsequent data rows.

This change checks sized indices against actual data rows, creates automatic indices with the same count, and filters DataFrame-owned labels alongside separator rows so each data row retains its original label. Explicit list/tuple indices contain one ID per data row; extra IDs previously accepted and silently discarded are now rejected.

The public regressions cover list, tuple, and iterator indices; single and multiple separators, including list-wrapped markers; shorter and longer indices; and DataFrame default/True/always indices. Literal table expectations verify both row labels and separators. Against the original source, 8 of the 11 cases fail and 3 controls pass.

Validation on Windows:

• Full suite with NumPy, Pandas, and wcwidth: 393 passed, 1 skipped; original existing tests: 382 passed, 1 skipped.
• Supported Python 3.10, 3.11, 3.12, 3.13, and 3.14 tox extra environments: each 393 passed, 1 skipped.
• All configured pre-commit hooks passed, including Ruff 0.15.4 checks and formatting.

This fix preserves existing iterator length behavior and does not address the separate trailing-separator rendering problem.

AI assistance was used for investigation, implementation, and review. The failures and checks above were executed locally, and the final source received independent reviews before submission.