Separate requested and effective excerpt ranges and add explicit EOF/clamp modes

[Widthdom]

Summary

Excerpt range normalization overwrites the requested range after applying context, line 0 is misdiagnosed as a missing start, and reading to EOF requires knowing the exact line count because overshoot is always fatal.

Audit IDs: Q-13, Q-32, and R-13. Reproduced with locally built cdidx v1.40.3 at 404db09489c305f17d94f37026e00564d3ac527a.

Existing issue relationship

No exact existing issue was found.

Expected behavior and PR scope

Preserve requested versus effective ranges, validate one-based inputs accurately, and add an explicit EOF and/or clamp mode without changing strict numeric defaults. These belong to one excerpt range parser/response PR.

Implementation guide for Codex 5.6 sol xhigh

• Model requested start/end, context expansion, effective clamped range, and total lines as separate values.
• Return a one-based invalid-range error for zero/negative inputs.
• Support a documented --end eof, --all, or explicit clamp flag; do not silently clamp strict numeric requests.
• Cover empty/one-line/newline/non-newline files, context at boundaries, overshoot, inline file:line, Unicode, and JSON recovery metadata.
• Add focused regression/contract tests, run the full net8.0 and net9.0 suite, and add the required bilingual changelog.d/unreleased/ fragment.

Acceptance criteria

• Requested 18–22 with context 2 reports requested 18–22 and effective 16–24.
• Line 0 receives the correct range taxonomy.
• Callers can explicitly read to EOF/clamp and strict behavior remains backward compatible.

[Widthdom]

Work started on branch fix-issue4877.