fix(tool): swap inverted file_read line range instead of erroring

[seamnniel]

Summary

file_read hard-errors when the model passes end_line < start_line:

file_read failed: invalid line range: start_line 430 is greater than end_line 120

This is a common model slip — the model derives start_line from the old side of a diff hunk while end_line comes from the new file's length. One bad call fails the whole file's review item (observed: 6/6 items in a review failed this way).

Fix: swap the inverted range and serve it. The response still reports Total lines, so the model can recover if the swap was not its intent. Updated TestExecute_InvalidLineRange to TestExecute_SwappedLineRange asserting the swapped range returns content.

Validation

• go test ./internal/tool/ — pass
• make check — pass
• Patched binary built and run locally: ocr llm test and ocr review --commit <sha> -p work.

Disclosure

This change was developed with AI assistance (Devin / SWE-2) per the project's agent guidelines; the change was reviewed and tested before submission.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[github-actions]

✅ OpenCodeReview: Review complete: 0 finding(s) across 1 selected item(s).

[PrayWithYou]

need CLA