fix: correct stale version numbers and update version-increment check…

[djkees]

Summary

Corrects stale version numbers (3.0) in two doc-generator config files that were left behind at previous releases, and closes the gap in the version-increment checklist that let this happen.

Note: this change has no visible effect on the currently published docs site — see the "Investigated impact" notes below.

Changes

• project.md: version: 3.0 → 3.3.2 (FORD doc-generator metadata)
• Doxyfile: PROJECT_NUMBER = 3.0 → 3.3.2 (Doxygen config)
• .codex/skills/version-increment/SKILL.md: added both files to the known hard-coded version locations list

Investigated impact before fixing:

• project.md/FORD: not referenced anywhere in the repo outside itself — appears unused by the current docs pipeline (docs.yml only invokes Doxygen + Sphinx, never ford). I corrected the value here in case FORD is still intended for future use, but if it's actually dead weight, happy to open a follow-up PR removing project.md entirely instead — just let me know which you'd prefer.
• Doxyfile: actively used — invoked by docs.yml in CI, and referenced in developer_guide.rst/faq.rst/installation.rst for local dev workflows — but PROJECT_NUMBER specifically feeds the intermediate Doxygen→Breathe→Sphinx pipeline and doesn't appear to render on the published site. The live docs' correct version comes from docs/source/conf.py, which was already accurate.

Testing

• ctest not applicable — no solver/numerics changes.
• Ran doxygen Doxyfile directly to confirm the corrected value doesn't break XML generation — completed successfully (one pre-existing, unrelated DOT_MULTI_TARGETS deprecation warning, unaffected by this change).
• project.md and SKILL.md have no associated build/test step (confirmed via repo-wide text search for each filename, excluding build artifacts).

Compatibility / Numerical behavior

• No expected changes to numerical results

---

I used Claude to help investigate this repo's docs pipeline and draft this description; the findings above were verified directly against the source (grep searches, reading docs.yml, running doxygen Doxyfile locally, and inspecting the generated XML/HTML output) rather than taken on faith.