fix(indeedhub): frontend health = tcp:7777 not http GET / (stops reconcile churn)

On the loaded .198 the frontend churned (created → "unhealthy" → reconciler
recreates → loop). The http health check fetched / through nginx (SPA +
sub_filter) and false-failed under node load; the reconciler then treated the
frontend as wedged and recreated it. nginx binds 7777 at startup, so a tcp
liveness check passes immediately and stays green under load while still
catching a real "nginx not listening" failure. Generous retries/start_period.

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

apps/indeedhub/manifest.yml

@@ -67,14 +67,18 @@ app:
       - exec: ["sh", "-c", "grep -q nostr-provider /etc/nginx/conf.d/default.conf || sed -i 's#</head>#<script src=\"/nostr-provider.js\"></script></head>#' /etc/nginx/conf.d/default.conf"]
       - exec: ["nginx", "-s", "reload"]
 
+  # TCP liveness on the nginx port, NOT an http GET of /. nginx binds 7777 at
+  # startup (before workers), so this passes immediately and stays green under
+  # load. An http check of / runs the SPA + sub_filter and false-fails when the
+  # node is busy → the reconciler then treats the frontend as wedged and
+  # recreates it in a loop (observed churning the frontend on the loaded .198).
   health_check:
-    type: http
-    endpoint: http://localhost:7777
-    path: /
+    type: tcp
+    endpoint: localhost:7777
     interval: 30s
-    timeout: 10s
-    retries: 3
-    start_period: 40s
+    timeout: 5s
+    retries: 5
+    start_period: 30s
 
   interfaces:
     main: