docs(plan): the dispatch document still said "ready to dispatch" with ten of its items shipped

[wshallwshall]

SCHEDULABLE-BACKLOG-MULTISESSION-PLAN.md is read by sessions deciding what to build. That is what makes this worse than an ordinary stale line: it does not misinform a reader about a fact, it misroutes work.

The failure is already on record for this exact file — a session dispatched from it this morning and inherited a superseded §D RULE 1, whose ownership clause the owner amended on 2026-08-04.

Verified before writing, not transcribed

• All ten claimed-shipped items are closed or archived: backlog: file #1060, alloc.ps1 records the caller's cwd as the owning worktree #233, backlog: amend #1008 -- built through a closed demand-gate; stays OPEN #322, backlog: close #214 and #249 on owner-delegated rulings (declined, not shipped) #324, fix(ci+test): the DAST closure guard matched inside INCOMPLETE, and the doc-guard list omitted it #326, backlog: file #1221 -- a doc lint crashes printing a hit that contains a glyph #330, backlog: file #1224 (upload.prune names the wrong actor) and #1225 (presets keyed on a reassignable name) #334, fix(test): pin the RecursionError contract without a real stack overflow (#1222) -- unreds main #335, backlog: amend #1225 -- the WRITE key at app.py:4054, and drop the three-site enumeration #336, ledger: file BACKLOG #1229 -- the gate's double-quote-first blanking straddles and hides live code #341, ci(caps): re-derive the ubuntu anchor and size both caps to it (BACKLOG #1096) #347
• Ledger is 108 open
• 26 open items are numbered ≥1000 and were filed after the plan existed — several, including #1014 and #1025-#1027, produced by the very lanes this plan dispatched

No running completion count, deliberately

Per CLAUDE.md §11 a count embedded in prose is a liability — it would go stale the moment the next lane lands, which is exactly how the line being fixed here got wrong.

Stated instead: the authoring baseline as fixed history (92 open items at authoring), a dated snapshot of what has shipped, and a pointer to the ledger as the live record. Same reasoning as the docs/SECURITY.md PHI-route rewording in #204 — stop asserting a count rather than correct one.

The planning gap is recorded, not closed

77 open items have no wave assignment and file contention has not been re-derived across the combined set, so scheduling from this document alone would both miss work and under-count collisions. Re-planning is a genuine multi-agent pass and an owner call — flagged as owed, not attempted here.

⚠️ This file sits inside docs/releases/, which ADR 0160 Phase 1 proposes removing from the public repo. Fixing it anyway: Phase 1 is unratified, and a dispatch document that misroutes work does so today. If Phase 1 lands, nothing is lost.

Docs-only, so #202's doc-guard leg runs on it — the second exercise of that step.