fix: increase postgres healthcheck start_period for first-boot init

On first deploy postgres must initialize the data directory before it
can accept connections. 10s start_period was too tight; bumped to 30s
with 10 retries to give it enough runway on slower hosts.

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

docker-compose.yml

@@ -50,8 +50,8 @@ services:
       test: ["CMD-SHELL", "pg_isready -U hermes -d hermes"]
       interval: 10s
       timeout: 5s
-      retries: 5
-      start_period: 10s
+      retries: 10
+      start_period: 30s
 
   hermes-control-plane:
     build: *hermes-build