test: scan the whole repo for stale doc paths, not just the package

[Gabeyocum28]

What does this PR do?

Widens tests/test_doc_references.py (added in #295) so it scans every place outside docs/ that points users at a docs/…​.md page, not just src/openflight:

• src/openflight/**/*.py
• scripts/**/*.py and scripts/**/*.sh
• .github/**/*.yml and .github/**/*.md (issue templates, PR template, workflows)
• README.md, CONTRIBUTING.md, CLAUDE.md, AGENTS.md, ui/README.md

One parametrized test per reference asserts the file exists, with the source file and path in the assertion message. The scan roots are an explicit list so a new tree is added deliberately. A guard test asserts the scan found more than ten references so it can never pass vacuously. Links inside docs/ are deliberately not scanned: the Zensical build runs with --strict in CI and already fails on broken intra-docs links.

Also adjusts the #295 changelog entry to describe the wider scan.

It already caught one. The first CI run failed on CONTRIBUTING.md pointing at docs/CHANGELOG.md; the file is docs/changelog.md. macOS let that pass locally because its filesystem is case-insensitive, so the test now checks every path component against the parent's real directory listing and is case-exact everywhere. The link is fixed in this PR.

Why was this required?

#294 fixed eleven stale doc paths in scripts, the issue templates, and the top-level guides by hand, using a one-off grep. #295 fixed four more inside the package and added a test, but that test only guards src/. The next docs restructure would strand the scripts and templates again with nothing to catch it. This turns the one-off cleanup into a permanent CI guard.

Automated tests

This PR is a test. Red proof: run against the tree from just before #294 merged (b65b5f3), the widened test fails on all fifteen paths that #294 and #295 fixed (the eleven in scripts/templates/guides plus the four in src/). On current main it passes with 46 references checked (after the CONTRIBUTING fix).

Manual (human) testing

• Ran the widened test on a git worktree at b65b5f3 (pre-docs: fix doc paths left stale by the docs restructure #294): 15 failed, 32 passed, each failure naming the stale source file and path.
• Ran it on this branch after rebasing onto current main: 46 passed. Confirmed the collected ids include references from all scan roots (scripts/setup/setup.sh, .github/ISSUE_TEMPLATE/hardware_help.yml, README.md, CLAUDE.md, src/openflight/kld7/radc.py, etc.).
• Full suite: uv run pytest tests/ gives 1840 passed, 11 skipped, 13 failed. The 13 are test_camera_* and fail identically on main in my environment because cv2 is not installed; unrelated.
• ruff check and ruff format --check clean on the test file.

No UI changes, so no screenshots.

Checklist

• Single feature/fix — this PR is scoped to one thing with a clear story above
• Automated tests included — new or updated tests cover this change
• Manual testing described — I documented what I verified by hand above
• Python tests pass (uv run pytest tests/ -v)
• Pylint passes (uv run pylint src/openflight/ --fail-under=9) — no src/ changes
• Ruff passes (uv run ruff check src/openflight/)
• UI builds (cd ui && npm run build) — no UI changes
• UI lint passes (cd ui && npm run lint) — no UI changes
• Updated docs or CHANGELOG if needed
• No unrelated changes mixed in