Point production at the LAN MongoDB (192.168.1.7:27017)

Production uses the network's database server instead of a mongo service
inside the stack (the compose-managed one never served the API). Dev is
unchanged: backend/.env keeps localhost:27017.

- docker-compose.yml: MONGO_URI=mongodb://192.168.1.7:27017/mathew; drop
  the in-stack mongo service, its volume and the healthcheck gate
- deploy.yml: up -d --remove-orphans so yesterday's mongo container is
  retired when the new compose lands

docker-compose.yml

@@ -19,32 +19,14 @@ services:
   backend:
     image: git.aranroig.com/syndria98/mathew-backend:latest
     environment:
-      # The image carries no .env (it is gitignored), and a container's
-      # 127.0.0.1 is itself — with the default URI the API crashed on connect
-      # and nginx answered 502. Mongo is a service of this stack now.
-      MONGO_URI: mongodb://mongo:27017/mathew
+      # Production MongoDB — the LAN database server, not a container of this
+      # stack and not the container's own 127.0.0.1 (where nothing listens).
+      # Development keeps localhost:27017 via backend/.env.
+      MONGO_URI: mongodb://192.168.1.7:27017/mathew
       # Interpolated from /var/www/app/.env on the deploy host (never
       # committed, never overwritten by the pipeline's scp). Create it once:
       #   echo "JWT_SECRET=$(openssl rand -hex 32)" >> /var/www/app/.env
       # Left unset, the backend falls back to the dev key in the repo —
       # anyone could then forge sessions against the public site.
       JWT_SECRET: ${JWT_SECRET:-}
-    depends_on:
-      mongo:
-        condition: service_healthy
     restart: always
-
-  mongo:
-    image: mongo:7
-    volumes:
-      - mongo-data:/data/db
-    healthcheck:
-      test: ["CMD", "mongosh", "--quiet", "--eval", "db.adminCommand('ping')"]
-      interval: 10s
-      timeout: 5s
-      retries: 5
-      start_period: 20s
-    restart: always
-
-volumes:
-  mongo-data: