validate-sectionlisted

# Validate a paper subsection Run a focused AI validation pass on one subsection of the paper, following the ledger conventions in `research-kit/meta/validation_ledger.md` and the methodology in `research-kit/meta/ai_research_workflow.md` stage 8. Invoke with a section identifier: - Section slug as it appears in the paper HTML anchor (e.g. `enforcement-outcomes`). - Or a section number (e.g. `7.4`). - Or a heading phrase that matches a subsection title. Flags: - `--full` — ignore the incremental cache (`last_ai_pass_sha`) and run every applicable check over the whole section. Use after a major prose rewrite, when changing the check vocabulary, or when suspicious that a prior pass missed something. - `--all` — iterate over every entry in `build/paper/section_deps.json` and run the skill once per section. Intended for autonomous bulk passes. Each section's result is recorded independently; a failure on one section does not abort the loop (the failing section is reported and the loop continues). Prose-only sections (no macros, no figures, no tables) are skipped — recording nothing for them keeps the ledger uncluttered. ## File locations The validation ledger is **project memory** (which scripts have been verified) and lives in `docs/` alongside `findings.md`, `hypotheses.md`, and the rest of project state. Paper-build outputs stay in `paper/` and `build/paper/`. Discovery order — first hit wins, with backward compatibility for the historical `paper/`