fix: healthcheck nunca retorna 503 — evita 'no available server'

health.php:
- DB fallo → 'warning' (no 'unhealthy', no cierra tráfico Traefik)
- Redis ya era warning desde commit anterior

docker-compose.yml:
- timeout: 5s → 10s (DB externa puede tardar más)
- start_period: 40s → 60s (tiempo suficiente para that migrations corran)
- fallback: si health.php falla, intenta / raíz antes de marcar unhealthy

docker-compose.yml

@@ -33,11 +33,11 @@ services:
     networks:
       - whatsapp-network
     healthcheck:
-      test: ["CMD", "curl", "-f", "http://localhost/health.php"]
+      test: ["CMD-SHELL", "curl -sf http://localhost/health.php || curl -sf http://localhost/ || exit 1"]
       interval: 30s
-      timeout: 5s
+      timeout: 10s
       retries: 3
-      start_period: 40s
+      start_period: 60s
     logging:
       driver: "json-file"
       options: