docs(overhaul): record S5 c1 mandatory stop

docs/research/2026-09-01-overhaul/s5-consumers-material-closeout-packet.md

@@ -1,6 +1,6 @@
 # Campaign OVERHAUL v2 — S5 consumers, material, and closeout packet
 
-**Status:** S5-c1 CONTRACTED 2026-09-04; implementation not yet landed.
+**Status:** S5-c1 STOPPED 2026-09-04 after fix round 2; not landed.
 **Branch:** `claude/campaign-w-retail-frame-walk`.
 **Gate:** G4 remains unpassed. Nothing merges to `main` before G4.
 
@@ -274,3 +274,46 @@ row and do not change the three AP-241/AP-242/AP-243 assertions. Run the exact
 failed test first, then a fresh official hermetic lane under a new artifact
 name and the required InstalledDat lane. Preserve the original failed artifact
 and record both identities in the implementer return.
+
+## 9. S5-c1 mandatory stop after fix round 2 — 2026-09-04
+
+The reviewed scratch stack is intentionally **not landed**:
+
+- gate-correction contract `2bb565b0b`;
+- implementation `9aadc917a`;
+- retail fix-round-1 contract/implementation `afaebcad6` / `74a2e4c48`;
+- retail fix-round-2 contract/implementation `ed8200664` / `a28384d80`.
+
+The retail lens ultimately passed. It verified the paired executable/PDB bytes,
+the exact land-cell versus EnvCell virtual behavior, previous-completed-frame
+timing, x87/AP-116 edges, typed landscape handoff, and unchanged particle
+draw/queue/tick behavior. The production lens found no production-code or test
+defect and reproduced Release 0 warnings / 0 errors, Core VFX 111/111, App
+focused 146/146, allocation 1/1 at 0 B, AP-116 3/3, all six non-vacuous
+mutations, all source guards, and all three 30-entry artifact manifests with
+zero mismatch. Official artifacts remain in the scratch worktree:
+
+- `artifacts/s5-c1-hermetic-20260904`: 16,759 passed / 1 stale-row failure;
+- `artifacts/s5-c1-hermetic-corrected-20260904`: 16,760/16,760;
+- `artifacts/s5-c1-installed-20260904`: 385 passed / 10 documented failed /
+  1 documented skipped, with no new identity.
+
+The final production lens nevertheless failed on two correction-worthy
+documentation/evidence findings after both allowed fix rounds:
+
+1. `acdream-architecture.md` and AD-21 say a null-root frame publishes no
+   cell-visibility answer. Production actually completes
+   `hasCompletedView=true` with an empty landscape set; only login/portal carry
+   false. Scratch packet §11.1 states this correctly, so the implementation's
+   claim that architecture/register prose is truthful is false.
+2. The packet reports the official lanes but does not reconcile their recorded
+   `WorktreeDirty=true` bases (`00f134f9e` for the first lane and `2bb565b0b`
+   for the corrected/InstalledDat lanes) to implementation commit `9aadc917a`,
+   nor state that later commits are prose/comments only. The manifests prove
+   their 30 recorded evidence files, not the dirty source content.
+
+Correcting either finding would be fix round 3. The campaign rule therefore
+requires this stop: do not cherry-pick the scratch stack, do not run a graphical
+gate on it, and do not dispatch S5-c2. Resumption requires explicit owner
+direction on whether to authorize an evidence/documentation-only exception;
+history must remain intact either way. G3 stays provisional and G4 unpassed.