Land pipeline diagram set (FIG-1..FIG-4) as committed mermaid

[WilfordGrimley]

Summary

• FIG-1 (pipeline flow, every interception point) added as the opening figure of docs/identification-pipeline.md.
• FIG-2 (catalog funnel) added to docs/pipeline-fidelity-gate.md §6 "Verified data snapshot" — shape only, no absolute counts, per owner ruling 2026-07-25 (the live numbers stay in this page's existing tables, right below the figure).
• FIG-3a (shakedown timeline) added to docs/features/stage-e-operations.md's trip runbook, stamped 2026-07-25T01:18:57Z as a dated snapshot: trip envtrip-...-be6e5db9 acknowledged, trip envtrip-...-73e1eb6d open at the stamp (root cause since fixed by Fix Stage E concurrency cap: hold advisory lock on a dedicated connection #453, but a fix does not self-clear a trip).
• FIG-4a (trip lifecycle stateDiagram-v2) + FIG-4b (5-gate anatomy table, including the concurrency cap updated for Fix Stage E concurrency cap: hold advisory lock on a dedicated connection #453's dedicated-connection fix) added to the same page's "envelope model" section.
• docs/theory.md §10 gains a link to FIG-1 — no diagram (theory.md deliberately avoids repo-internal jargon).
• docs/documentation-process.md gets a durable note recording the deferred site-page decision (mermaid fences are wiki-only for now — marked has no mermaid support; three options recorded, owner sign-off needed for any of them).
• .github/wiki-publish-map.json gains a docs/pipeline-fidelity-gate.md → Pipeline-Fidelity-Gate wiki entry (wiki-only, no site target).

Owner rulings implemented

1. FIG-2 ships shape-only, no baked-in numbers — done, see caption in the diagram's own section.
2. FIG-3 is a dated post-mortem snapshot, not a live readout — stamped, both trip IDs named with their state at the stamp.
3. Site page: NOT NOW. No frontend mermaid dependency added; wiki-first; the deferred option is recorded in docs/documentation-process.md, not dropped.

Test plan

• python3 .github/scripts/docs_lint.py --strict → clean.
• .github/scripts/tests/test_publish_wiki_link_rewrite.py → 4 passed, 28 subtests.
• .github/scripts/tests/test_publish_readme.py and test_docs_lint.py → all passed.
• .github/scripts/publish_wiki.py run locally against a scratch dir → all pages generate with zero link-resolution errors; confirmed both new mermaid fences (FIG-1, FIG-2) survive the wiki transform verbatim.
• readme.md regenerated locally via publish_readme.py → no diff (parity holds).
• prettier pre-commit hook ran clean on the final commit.
• Mermaid rendering itself not screenshotted (no local mermaid renderer in this sandbox) — verified by hand-tracing node/edge syntax against Quorra's spec's own verbatim blocks, and by the wiki dry-run above confirming the fences pass through untouched; CI/GitHub's own renderer is the real check once this PR is live.

Merge-time checklist

• wiki: Identification-Pipeline needs its new FIG-1 opening figure (regenerates automatically on merge via the docs-wiki-publish workflow — no manual step, just confirm it ran).
• wiki: Pipeline-Fidelity-Gate is a newly published page (was not on the wiki before this PR) — confirm it appears after merge.
• wiki: stage-e-operations.md (FIG-3a/4a/4b) remains unpublished to the wiki, unchanged from before this PR — intentional, matches the existing HOLD-gated/operator-only posture for that page; flagging only so a reviewer doesn't read the absence as an oversight.

Open items (owner-only)

1. Site-page rendering for FIG-1/FIG-2 (mermaid on /guide) — three options recorded in docs/documentation-process.md, none implemented; needs owner sign-off before any of them.
2. FIG-3's quantitative load-vs-ceiling SVG chart (Quorra's D-7) was deliberately not committed — no asset-maintenance owner assigned yet, and the real tuning numbers it would plot don't exist until the phase-3 shakedown. Noted inline in FIG-3a's own section; can follow later.

🤖 Generated with Claude Code