Improve documentation image builds

[yawkat]

Closes #45

Summary

• canonicalize OpenSCAD PNG output so identical pixels produce identical files
• normalize the existing documentation images without changing their pixels
• verify just docs output in GitHub Actions using a checksum-pinned OpenSCAD development AppImage

Validation

• all 97 rewritten images are pixel-identical to their committed inputs
• differently compressed identical images canonicalize to identical bytes
• canonicalization is idempotent
• uv sync --frozen --offline
• just test
• headless PNG export with OpenSCAD AppImage 2026.07.16

[github-actions]

🚀 Preview deployed to https://cbbaa93a.gridflock-yawk-at.pages.dev

[Copilot]

Pull request overview

This PR improves reproducibility of documentation image generation by adding a PNG canonicalization step and a CI workflow that verifies just docs leaves committed docs images unchanged.

Changes:

• Add a Pillow-based canonicalize_png.py utility and run it during just docs image generation.
• Add a GitHub Actions workflow to run just docs with a checksum-pinned OpenSCAD AppImage and verify docs/images has no diffs.
• Add the pillow dependency to the Python project and lockfile.

Reviewed changes

Copilot reviewed 4 out of 102 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
uv.lock	Adds the locked Pillow dependency required for PNG canonicalization.
pyproject.toml	Declares pillow as a runtime dependency.
justfile	Runs canonicalization on generated docs images during just docs.
canonicalize_png.py	Introduces a deterministic PNG rewrite utility using Pillow.
.github/workflows/docs.yml	Adds CI verification that just docs output matches committed docs/images using a pinned OpenSCAD AppImage.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.