Add frontend performance metrics pipeline to Prometheus

Collect browser-side worker round-trips, computation times, main-thread
operations, and feature counts, batch them client-side, and expose as
Prometheus histograms via a new POST /api/perf endpoint.

frontend/src/services/perfCollector.ts

@@ -0,0 +1,46 @@
+interface PerfSample {
+  metric: string;
+  operation: string;
+  value: number;
+}
+
+const FLUSH_INTERVAL_MS = 30_000;
+let batch: PerfSample[] = [];
+let flushTimer: ReturnType<typeof setInterval> | null = null;
+
+export function record(metric: string, operation: string, value: number): void {
+  batch.push({ metric, operation, value });
+}
+
+function flush(): void {
+  if (batch.length === 0) return;
+  const blob = new Blob([JSON.stringify(batch)], { type: 'application/json' });
+  batch = [];
+
+  if (navigator.sendBeacon) {
+    navigator.sendBeacon('/api/perf', blob);
+  } else {
+    fetch('/api/perf', {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: blob,
+      keepalive: true,
+    }).catch(() => {});
+  }
+}
+
+export function startCollector(): void {
+  if (flushTimer) return;
+  flushTimer = setInterval(flush, FLUSH_INTERVAL_MS);
+  document.addEventListener('visibilitychange', () => {
+    if (document.visibilityState === 'hidden') flush();
+  });
+}
+
+export function stopCollector(): void {
+  if (flushTimer) {
+    clearInterval(flushTimer);
+    flushTimer = null;
+  }
+  flush();
+}