Wave-4 general review (002): two live-proven bugs — the forwarder's permanent death after one failed reconnect (the failure arm the gate's test never covered) and the tx enqueue/publish paths' missing pg_notify wake (F-1's root cause, engine-side) — plus a max_size:0 open-hang, two narrow robustness gaps, doc mismatches, and the decode-duplication smell; reviews renumbered 001/002 per the alk* numbering pattern (references updated)

docs/architecture/decisions/023-fourth-review-round.md

@@ -4,7 +4,7 @@
 
 Accepted (2026-10-08, Phase 1 — fourth review round, following the
 2026-10-08 waves-1–2 general review
-(`docs/reviews/2026-10-08-waves-1-2-general-review.md`). Amends
+(`docs/reviews/001-waves-1-2-general-review.md`). Amends
 [ADR-020](020-enqueue-opt-semantics-and-bridges.md) §4,
 [ADR-008](008-contract-v1-pinning.md) §5, and
 [ADR-012](012-forked-substrate-design.md) §4 in place — pre-implementation
@@ -306,7 +306,7 @@ and removes the tuning lever the config surface exists for.
 
 ## References
 
-- `docs/reviews/2026-10-08-waves-1-2-general-review.md` — the review
+- `docs/reviews/001-waves-1-2-general-review.md` — the review
   report that found all four (N-2, M-2, N-4, N-6; M-2's engine-layer
   resolution point recommended there is adopted *where*, while this
   ADR decides *what*, contract-side).