start: group Docker containers by compose project (collapsible stacks with running count, health dot, CPU/mem); collector captures compose labels

deploy/home-collector/home-collector.py

@@ -4,7 +4,7 @@
 Runs every minute as a systemd user timer on the server (home-collector.timer) and upserts
 two rows into home.snapshots with the limited `home_collector` role:
   host  CPU, RAM, swap, disk, load, uptime, Docker containers (state, health, restarts,
-        CPU/mem), systemd timer results (backups, disk-guard, scrapers) and service checks
+        CPU/mem, compose project/service), systemd timer results (backups, disk-guard, scrapers) and service checks
   ntfy  the most recent messages from the owner's ntfy topics
 
 The app container has no Docker socket and never gets one; this script uses the host's
@@ -154,6 +154,7 @@ def containers():
     for c in info:
         st = c.get('State', {})
         name = c.get('Name', '').lstrip('/')
+        labels = c.get('Config', {}).get('Labels') or {}
         out.append({
             'name': name,
             'image': c.get('Config', {}).get('Image', '').split('@')[0][:80],
@@ -164,6 +165,9 @@ def containers():
             'startedAt': st.get('StartedAt'),
             'cpu': stats.get(name, {}).get('cpu'),
             'mem': stats.get(name, {}).get('mem'),
+            # Compose project/service, so /start can group containers by stack.
+            'project': labels.get('com.docker.compose.project'),
+            'service': labels.get('com.docker.compose.service'),
         })
     out.sort(key=lambda c: c['name'])
     return out