Adds OCR evaluation script / Fixes WER calculation

[tomaroberts]

This PR:

• adds OCR evaluation script, run on 20 records
• fixes WER metric code

[tomaroberts]

TODO: this PR branched off old branch prior to analysis folder being moved into src/pyonb/analysis – hence conflicts.

[tomaroberts]

Re-run analysis today – results more sensible now.

[Copilot]

Pull Request Overview

This PR adds an OCR evaluation script and fixes the WER (Word Error Rate) calculation method. The changes improve the accuracy of text comparison metrics and make the evaluation functionality more modular.

• Replaces Levenshtein string-based WER calculation with a proper word-based dynamic programming implementation
• Refactors OCR evaluation script to be more modular by extracting a run() function
• Adds disk space management to GitHub Actions workflow to handle build requirements

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.

File	Description
src/pyonb/analysis/metrics.py	Fixes WER calculation with proper dynamic programming and updates metric documentation
src/pyonb/analysis/eval_ocr.py	Refactors evaluation script to be more modular and adds file encoding parameter
.github/workflows/tests.yml	Adds disk space cleanup step to prevent build failures

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[p-j-smith]

thanks for fixing this @tomaroberts!