feat(edge): community IP blocklist (Spamhaus DROP) at the gateway

Refuse a client whose IP is in a curated CIDR feed with 403 in the same
abuseGuard, before the fail2ban ban. Prod-only (keys by real client IP), off by
default.

- ratelimit.Blocklist: a sorted-range IPv4 matcher (binary search) with an
  allowlist checked first; ParseDROP reads the feed; ApplyRefresh keeps the
  last-good feed on a transient fetch failure and drops it fail-open once stale
  (better to under-block than block a legitimate client on a frozen feed). A
  separate static CIDR set, not the per-IP fail2ban store.
- gateway: a refresher goroutine re-fetches every few hours (bounded fetch + size
  cap); config GATEWAY_BLOCKLIST_{ENABLED,URL,ALLOW,REFRESH,MAX_STALENESS}.
  IPv6 is not matched (a v6 client is still covered by fail2ban / the honeypot).
- observability: gateway_blocklist_blocked_total + entries/age gauges; a Grafana
  alert warns before the feed is dropped; service-overview panels.
- deploy: compose env + write-prod-env.sh + prod-deploy/rollback wiring (opt-in
  via PROD_ vars).
- docs (ARCHITECTURE, deploy/README); unit tests (match, allowlist, IPv6 skip,
  parser, fault-tolerance) + a 403 abuse-guard integration test.

docs/ARCHITECTURE.md

@@ -1206,6 +1206,18 @@ link — misses the event; while an add-email confirmation is pending the client
   (`POST /api/v1/internal/bans/sync`, network-trusted like the rejection report) and applies
   the operator unbans the response returns, so a manual unban takes effect within the sync
   interval.
+- **Community IP blocklist (prod-only):** with `GATEWAY_BLOCKLIST_ENABLED` set, the gateway also
+  refuses a client whose IP is in a curated CIDR feed (Spamhaus DROP, `GATEWAY_BLOCKLIST_URL`) with
+  **403** in the same `abuseGuard`, before the fail2ban list. A background refresher re-fetches the
+  feed every few hours (bounded fetch + size cap) into a sorted-range matcher (`ratelimit.Blocklist`,
+  binary search, IPv4 only — an IPv6 client is never blocked here). It is **fault-tolerant and
+  fail-open**: a failed fetch keeps the last-good feed, and once the feed is older than
+  `GATEWAY_BLOCKLIST_MAX_STALENESS` it is **dropped** rather than block a legitimate client on a
+  frozen list; a `GATEWAY_BLOCKLIST_ALLOW` allowlist (own infra, monitoring) is never blocked. Off by
+  default and prod-only for the same real-client-IP reason as the ban. It is a **separate** static
+  CIDR set, not the per-IP fail2ban store (a bulk CIDR feed cannot expand into per-IP entries).
+  Observability: `gateway_blocklist_blocked_total`, the `gateway_blocklist_entries` size gauge and the
+  `gateway_blocklist_age_seconds` staleness gauge (a Grafana alert warns before the feed is dropped).
 - Unauthenticated `GET /healthz` (liveness) and `GET /readyz` (readiness — the
   database answers a bounded ping and the session cache is warmed).
 - The backend serves a **second listener** — a gRPC server