fix: change CouchDB service from headless to ClusterIP for DNS resolution

Headless services (clusterIP: None) don't get DNS entries for the service name itself,
only for individual pods. This was causing DNS resolution failures for the backend
trying to connect to adopt-a-street-couchdb.

Since we only have 1 replica, a regular ClusterIP service works better and provides
proper DNS resolution.

Fixes:
- Backend can now resolve adopt-a-street-couchdb DNS name
- CouchDB connection is stable
- Health endpoint returns connected status

Deployment status:
- Backend: 1/1 Ready, healthy, connected to CouchDB
- Frontend: 1/1 Ready, serving nginx
- CouchDB: 1/1 Ready, StatefulSet with 10Gi storage
- Ingress: Routing working at 192.168.153.241

🤖 Generated with AI Assistant

Co-Authored-By: AI Assistant <noreply@ai-assistant.com>

deploy/k8s/couchdb-statefulset.yaml

@@ -6,7 +6,7 @@ metadata:
   labels:
     app: couchdb
 spec:
-  clusterIP: None  # Headless service for StatefulSet
+  type: ClusterIP  # Regular ClusterIP service (not headless)
   selector:
     app: couchdb
   ports: