feat(onboarding): template-driven scaffold + register prompts/eval in matrix

[arnaudlh]

Summary

End-to-end overhaul of the /git-ape-onboarding flow plus the supporting packaging, instructions, eval registration, and regenerated docs. Five themed commits, each independently revertable:

#	Commit	Why
1	chore(models): rename gpt-5-codex to gpt-5.3-codex	waza catalog renamed; aligns manifest tiers + bench prompt defaults
2	feat(onboarding): replace exampleyml stubs with template-driven scaffold	Move workflow scaffolds out of .github/workflows/ (where the .exampleyml hack lived) into .github/skills/git-ape-onboarding/templates/; ship sync scripts; add eval suite + CI parity check; register the eval in the pilot tier
3	feat(extension): register prompt files in VSIX and tighten .vscodeignore	plugin.json now declares prompts:; package.template.json registers all 9 prompts as chatPromptFiles; .vscodeignore sheds dev-only .github subtrees from the published VSIX
4	docs(instructions): switch deploy/destroy guidance to Azure Deployment Stacks	Copilot-instructions caught up with the actual templates: az stack sub create/delete --action-on-unmanage deleteAll instead of az deployment sub + az group delete (see #30)
5	docs(website): regenerate for templated workflows and prompt assets	Generator now scans both .github/workflows/ and the skill templates directory, tags scaffolded workflows with a Docusaurus admonition

Key change: .exampleyml is gone

Old shape (this repo's .github/workflows/):

git-ape-plan.exampleyml
git-ape-deploy.exampleyml
git-ape-destroy.exampleyml
git-ape-verify.exampleyml

.exampleyml was a workaround so GitHub Actions wouldn't auto-load the scaffolds.

New shape (.github/skills/git-ape-onboarding/templates/workflows/):

git-ape-plan.yml
git-ape-deploy.yml
git-ape-destroy.yml
git-ape-verify.yml
git-ape-drift.md          # agentic workflow source
git-ape-drift.lock.yml    # compiled lockfile

The path is no longer .github/workflows/ so the workaround isn't needed. /git-ape-onboarding copies these into the target repo with skip-on-collision so customized workflows are never overwritten.

Eval registration

Adds git-ape-onboarding to pilot tier in .github/evals/manifest.yaml — matches its prior 4-model bench coverage. The eval ships 4 tasks: positive-first-time-setup, positive-multi-env, positive-skip-on-collision, negative-storage-comparison. Closes part of #93.

Verification done

• actionlint clean on the new git-ape-onboarding-template-check.yml
• node scripts/generate-docs.js re-runs with no further drift
• Eval YAML files parse
• All shell scripts under scripts/ carry the executable bit

Dependency

Depends on #140 (LLM-as-judge → claude-opus-4.7). This PR is mergeable independently — if it lands first, the new git-ape-onboarding eval will run with whatever judge is pinned at merge time, then automatically pick up the opus judge once #140 merges. Prefer to merge #140 first to avoid a mixed-judge snapshot.

Risk

Medium-low.

• Workflow templates are new files in a non-loaded path, no production CI impact.
• The 4 .exampleyml deletions are the only destructive change in this repo; their content is preserved verbatim in the templates directory.
• Eval registration adds 4 new task runs to the next waza-evals matrix dispatch (pilot × 4 models = 16 legs). Quota cost: ~equivalent to prereq-check baseline.

[arnaudlh]

Superseded by #142, which is the same commits (8d5b9cf5..e66e2ecc) pushed to a branch on Azure/git-ape itself so the waza eval matrix can read COPILOT_GITHUB_TOKEN. Fork PRs were skipping all eval legs at preflight. Closing this one to keep review in a single thread.