feat(molecules): add refiners-update molecule

[ShuratCode]

What

Adds refiners-update — the update-mode counterpart to lattice-init. After a significant change (architecture shift, language switch, new domain rules), it scans .lattice/standards/ for existing refiner outputs, asks what changed, routes each affected standard to that refiner's existing revise mode, and records a git-native Last updated: DATE — reason note on each revised doc.

Reframe vs. the issue

Issue #30 proposed a refiner called refiners-update that bumps a context.yaml to "v2". Two adjustments after reading the codebase:

• Tier: molecule, not refiner. It orchestrates the existing refiners (each already has a revise / start-fresh / add path) rather than running its own interview — exactly the shape of lattice-init. Refiners are single-interview producers; this coordinates several.
• Versioning: git-native, no "v2". Lattice has no content-versioning (config.yaml's version: 1 is the config schema version, not per-standard content). Rather than invent one, history stays in git log and each revised doc carries a one-line change note.

Changes

• skills/molecules/refiners-update/SKILL.md — the molecule (composes framework:knowledge-priming)
• Docs sync: inventory row + in-depth section in docs/how-it-works.md, dropdown entry in .github/ISSUE_TEMPLATE/bug_report.yml, counts in PROJECT.md / README.md (26 → 27 skills, 8 → 9 molecules)
• Regenerated Codex plugin package (plugins/lattice/skills/refiners-update/)

Testing

• ./tools/install.sh <dir> → "Installed 27 skills"; refiners-update flattens with valid frontmatter
• skill-validate: PASS (0 errors)
• skill-review (3 personas): 1 critical + 2 observations found and fixed — the execution/control model for the change-note append, a learning→standard mapping row, and a maintainer sync note

Closes #30