fix(#355 review round-2 F9-F11): register-gate test + shutdown idle-close + DB-path metrics gate

- F10 [stability]: closeMetricsServer() now calls server.closeIdleConnections()
  + server.unref() after server.close(). server.close()'s callback doesn't fire
  until keep-alive sockets drain, and the scraper (VictoriaMetrics/vmagent) holds
  an idle keep-alive socket — so onModuleDestroy's awaited close would hang until
  the scraper disconnects or the orchestrator SIGKILLs on the kill-grace window.
  closeIdleConnections() drops idle keep-alive sockets so shutdown completes
  immediately (Node 22, per the Dockerfile base).
- F9 [test]: client-telemetry.module.spec.ts pins the E1=B register() gate — the
  core of the "public endpoint OFF by default" decision: flag unset / any non-
  "true" value ("false"/""/"0"/…) → empty controllers+providers (route absent);
  "true"/"TRUE" → registers VitalsController + VitalsService. A flag-inversion or
  truthiness regression that reopened the anonymous disk-fill surface now fails.
- F11 [regression/perf]: the db_query_duration_seconds token work (firstSqlToken
  regex + Set lookup) is now gated on isMetricsEnabled() in database.module.ts, so
  a non-metrics deployment pays NOTHING per query (previously observeDbQuery
  no-op'd but the token was still computed on every query). Also hoisted the
  13-element known-token Set to a module const (KNOWN_SQL_TOKENS) so it's built
  once, not per query.

Gate: server tsc 0; metrics + vitals + client-telemetry suites pass (incl. the
new register-gate test).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

apps/server/src/integrations/metrics/metrics.constants.ts

@@ -30,28 +30,32 @@ export const JOB_BUCKETS = [
  * the full query text would blow up label cardinality; the leading keyword is a
  * finite set. Unknown/empty queries collapse to `other`.
  */
+// The bounded set of SQL leading keywords used as db_query_duration_seconds
+// labels. Module-const so it is built ONCE, not per query (this runs on every DB
+// query when metrics are enabled).
+const KNOWN_SQL_TOKENS = new Set([
+  'select',
+  'insert',
+  'update',
+  'delete',
+  'with',
+  'begin',
+  'commit',
+  'rollback',
+  'alter',
+  'create',
+  'drop',
+  'truncate',
+  'explain',
+]);
+
 export function firstSqlToken(sql: string | undefined): string {
   if (!sql) return 'other';
   // Skip leading whitespace / comments and grab the first word.
   const match = /^[\s(]*([a-zA-Z]+)/.exec(sql);
   if (!match) return 'other';
   const token = match[1].toLowerCase();
-  const known = new Set([
-    'select',
-    'insert',
-    'update',
-    'delete',
-    'with',
-    'begin',
-    'commit',
-    'rollback',
-    'alter',
-    'create',
-    'drop',
-    'truncate',
-    'explain',
-  ]);
-  return known.has(token) ? token : 'other';
+  return KNOWN_SQL_TOKENS.has(token) ? token : 'other';
 }
 
 /**

apps/server/src/integrations/metrics/metrics.server.ts

@@ -73,5 +73,14 @@ export function closeMetricsServer(): Promise<void> {
   if (!server) return Promise.resolve();
   return new Promise((resolve) => {
     server.close(() => resolve());
+    // server.close() stops accepting NEW connections but its callback does not
+    // fire until existing keep-alive sockets drain. The scraper (VictoriaMetrics/
+    // vmagent) holds an idle HTTP keep-alive socket, so without this the callback
+    // — and thus shutdown — would hang until the scraper disconnects or the
+    // orchestrator escalates to SIGKILL on the kill-grace window. Force-close idle
+    // keep-alive sockets so close() completes immediately, and unref so this
+    // server never keeps the event loop alive on its own.
+    server.closeIdleConnections();
+    server.unref();
   });
 }