fix(compose): drop service_healthy, use service_started + sleep

pg_isready healthchecks broken server-wide (root_db, hoppscotch-db also
unhealthy). Remove healthcheck block, depend on service_started only.
migrate sleeps 5s to give postgres time to accept connections.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

docker-compose.yml

@@ -8,19 +8,11 @@ services:
       POSTGRES_PASSWORD: ${POSTGRES_PASSWORD:-changeme}
     volumes:
       - postgres_data:/var/lib/postgresql/data
-    healthcheck:
-      test: ["CMD", "pg_isready", "-U", "grow", "-d", "grow", "-h", "127.0.0.1"]
-      interval: 5s
-      timeout: 5s
-      retries: 20
-      start_period: 15s
-
   app:
     build: .
     restart: unless-stopped
     depends_on:
-      db:
+      - db
-        condition: service_healthy
     ports:
       - "${PORT:-3000}:3000"
     environment:
@@ -43,10 +35,9 @@ services:
     build:
       context: .
       target: builder
-    command: node_modules/.bin/prisma db push
+    command: sh -c "sleep 5 && node_modules/.bin/prisma db push"
     depends_on:
-      db:
+      - db
-        condition: service_healthy
     environment:
       DATABASE_URL: "postgresql://grow:${POSTGRES_PASSWORD}@db:5432/grow"
     restart: "no"