feat: deliver source-built documentation

[Pluze]

Why

Generated site/ output duplicated the Markdown and MATLAB-help sources, made documentation edits produce large mechanical diffs, and allowed Pages to publish a stale committed tree. The expanded MATLAB compatibility matrix also serialized the longest independent profiles and repeated the complete expensive matrix after an already-required protected PR run.

What changed

• Removed generated site/ files from Git tracking and ignored local renders.
• Made GitHub Pages build and deploy documentation from the exact main source.
• Made docsCheck compare two independent temporary renders by default.
• Made Launcher documentation online-first, with explicit local lookup and an Open Online-default prompt to generate missing local documentation.
• Split only the measured latest Windows and macOS critical paths into Core and Hidden GUI shards.
• Kept the complete MATLAB matrix on PRs while reducing the exact post-merge main run to policy and aggregate-gate evidence.
• Updated launcher version/history, manuals, repository policy, summary generation, and regression tests.

Evidence

python -m unittest discover -s .github/scripts -p 'test_*.py'
16 passed

matlab -batch "buildtool changedFast"
96 tests, 0 failed, 0 incomplete

buildtool docsCheck
375 deterministic generated files

LauncherDispatchSpec bounded closure
40 tests, 0 failed

TestArchitectureSpec focused run
14 tests, 0 failed

maintainLabKitDocLinks
243 Markdown files checked; 0 unresolved links

Clean render into a missing output directory
223 narrative pages, 147 API pages, 375 files

Online Batch Crop documentation route
HTTP 200

git ls-files site
0 tracked files

Risks and follow-up

The external browser handoff and native uiconfirm presentation were not interactively exercised; automated hidden-GUI checks cover callback routing but do not prove native dialog appearance. GitHub Pages deployment and the new seven-job MATLAB shard shape will be verified by this PR and its exact merged-main workflows. No new runtime dependency or scientific behavior is introduced.

Author confirmation

• The final diff matches the stated scope and contains no unrelated change.
• The evidence above covers required local and manual validation; every
  remaining unverified boundary is explicit.
• The diff contains only synthetic or generic public data, with no private
  App details, lab identifiers, local paths, or recognizable sample values.