Tagged v1/v2 Coexistence — Merge doc/v2/ and Add Version Markers

The v1 → v2 doc migration began by isolating new content under doc/v2/. In practice the split has broken down: v2-shaped docs now live in three places (doc/v2/, projects/*/modeling/component_overview.org, and a long tail), v1 docs keep getting edited under their original paths, and there is no single source of truth per topic. Reviewers cannot tell which version is canonical, and tooling cannot measure progress.

The new model:

1. Every doc carries #+version: 1 or #+version: 2.
2. doc/v2/ is merged into the main doc/ tree — no separate subdir.
3. Touch-it-to-port-it: editing a v1 doc means full-rewriting it to v2 (verifiable by the v2 frontmatter contract). Small fixes still trigger the rewrite — that is the intentional pressure.
4. A tracking script enumerates remaining v1 docs so we know how far we have to go.

• Every doc-shaped .org file has #+version: 1 or #+version: 2.
• doc/v2/ no longer exists; its contents live at their new top-level paths inside doc/.
• v1 sprint backlogs 01–16 plus version_zero.org are deleted (v2 trees already cover them).
• sprint_backlog_17.org stays tagged as v1 (active sprint, no v2 form yet).
• validate_docs.sh enforces #+version: on every doc that carries #+type:.
• projects/ores.codegen/scripts/doc_version_audit.py lists remaining v1 docs and reports the % migrated.
• CLAUDE.md codifies the touch-it-to-port-it rule and points at the audit script.
• All scaffolds (v2-doc-author skill, mustache templates, doc_generator.md) inject #+version: 2 and reference the new doc paths.

Confirmed-migrated to doc/v2/versions/v0/sprint_NN/...:

Delete these v1 .org files (17):

• doc/agile/v0/sprint_backlog_01.org through _16.org
• doc/agile/v0/version_zero.org (v2 equivalent: doc/v2/versions/v0/version.org)

Delete the 49 auto-generated chart PNGs (only referenced from the deleted org files):

• sprint_backlog_NN_stories.png, _tags.png, _stories_pie_sorted.png, _plan.png, _resources.png for NN = 01..16.

Delete the two release-notes markdowns (zero inbound references confirmed):

• sprint_backlog_15_release_notes.md, sprint_backlog_16_release_notes.md

Keep:

• sprint_backlog_17.org — active sprint, no v2 tree yet. Remains tagged #+version: 1 from Phase 1.
• 10 misc UI mockup PNGs (dashboard_ideas.png, figma_user_dashboard.jpeg, etc.) — still referenced from doc/agile/product_backlog.org (v1, lazy-port).

No filename collisions found between doc/v2/{knowledge,recipes}/ and doc/{knowledge,recipes}/ (verified by recursive comm).

• projects/ores.codegen/validate_docs.py: add a MISSING_VERSION check requiring #+version: 1|2 on every file with #+type:. Currently it only validates component_overview.org files; extend the scope or wire --check of doc_version_audit.py into validate_docs.sh so both run.
• CLAUDE.md: add a "Documentation versions" section explaining the marker, the touch-it-to-port-it rule, and pointing at doc_version_audit.py --list-v1.
• Scaffolds inject #+version: 2:
  • doc/llm/skills/v2-doc-author/SKILL.org (and any sub-templates)
  • projects/ores.codegen/library/templates/doc_skill.org.mustache
  • projects/ores.codegen/docs/doc_generator.md (update example outputs to show the new header)
• Re-run python projects/ores.codegen/scripts/doc_version_audit.py --check — must pass.

• Push feature/doc-version-marker, open the PR.
• In the body, include the audit summary (vN / total = pp%) so reviewers can see the migration runway.

1. sprint_backlog_17.org: confirmed plan keeps it as v1 (active sprint). If you'd rather it be deleted now and the v2 sprint_17 tree authored in this PR, that is a much bigger scope — flag it.
2. Pure-scaffold allowlist: the audit script will exempt doc/doc.org, readme.org, index.org, and any other file flagged as having no body and no #+type:. If you want even those tagged, the rule is trivially relaxed.

• projects/ores.codegen/scripts/doc_version_audit.py (new)
• projects/ores.codegen/validate_docs.py (extend)
• projects/ores.codegen/validate_docs.sh (wire in the audit check)
• .build-site.el (nav link)
• readme.org (path text)
• index.org (gets #+version: 1)
• CLAUDE.md (touch-to-port rule)
• doc/llm/skills/v2-doc-author/SKILL.org
• projects/ores.codegen/library/templates/doc_skill.org.mustache
• projects/ores.codegen/docs/doc_generator.md

Plus mass-stamping ~1000 .org files across the repo (via the script in Phase 1).

End-to-end checks before opening the PR:

1. python projects/ores.codegen/scripts/doc_version_audit.py — table reports 0 untagged docs.
2. python projects/ores.codegen/scripts/doc_version_audit.py --check — exit 0.
3. ./projects/ores.codegen/validate_docs.sh — passes (includes MISSING_VERSION check).
4. emacs -Q --script .build-site.el — build succeeds, no broken [[file:]] links.
5. ./build/scripts/serve-site.sh — browse:
   • /OreStudio/index.html
   • /OreStudio/readme.html
   • /OreStudio/doc/compass.html (formerly under /doc/v2/)
   • A sample sprint page like /OreStudio/doc/versions/v0/sprint_16/...html
   • A sample component overview from projects/ores.qt.trading/modeling/component_overview.html
6. git grep 'doc/v2/' returns no hits outside this plan file.
7. Output of doc_version_audit.py --list-v1 | wc -l quoted in the PR body as the migration runway baseline.