docs(ms-ai-architect): R11 §10 pilot-tall retalt etter R7.5-komplettering — 21/178 → 24/202 [skip-docs]

Pilot-utvalget i §10 ble målt mot 222-record-ledgeren. R7.5 er nå komplett (243
records, +82 flagg), så tallet var utdatert i det øyeblikket 8b270fb landet.

Designregelen er uendret — tetteste ≥7-utvalg. Kun tellingen flyttet seg:
- 21 filer / 178 not_grounded → 24 filer / 202
- Nye i utvalget: semantic-caching-patterns, small-language-models-economics,
  vector-storage-cost-optimization

Presiserer også terskel-definisjonen, som ikke sto eksplisitt: den teller
not_grounded alene, ikke source_silent. På alle flagg ville utvalget vært
39 filer / 352 — en dobling, og en reell felle for pilotøkten.

Karakteristikken holder fortsatt: 202/776 = 26 % av volumet i 24/229 = 10 %
av filene.

docs/r11-tiered-fix-design.md

@@ -182,9 +182,17 @@ full corpus pass to the plan.
 
 ## 10. Pilot and acceptance criterion
 
-Before any scaling, run the classifier and the O1 driver against the **21 files
-carrying ≥7 flags (178 flags)** — a quarter of the volume in a tenth of the
-files, and the densest available sample.
+Before any scaling, run the classifier and the O1 driver against the **24 files
+carrying ≥7 `not_grounded` flags (202 flags)** — a quarter of the volume in a
+tenth of the files, and the densest available sample.
+
+> Recount 2026-08-03 after R7.5 completed (ledger 222 → 243 records). The rule is
+> unchanged — densest ≥7 sample — only the count moved: 21 files / 178 flags was
+> measured against the 222-record ledger, before the last 21 R7.5 files were
+> ingested. Three files entered the sample (`semantic-caching-patterns.md`,
+> `small-language-models-economics.md`, `vector-storage-cost-optimization.md`).
+> The threshold counts `not_grounded` only, not `source_silent`; on all-flags it
+> would be 39 files / 352.
 
 Measure and record: