Reduce pool size and deployment surge to prevent PG connection exhaustion (#94)

PG max_connections=100 (97 usable). With pool_size=15 and maxSurge=100%,
rolling deploys briefly run 4 pods × 15 = 60 pool connections plus Oban
notifiers, exceeding the limit and causing too_many_connections errors
that cascade into ssl recv: closed timeouts.

- pool_size: 15 → 10 (steady state: 2×10=20, deploy: 3×10=30)
- maxSurge: 100% → 1 (max 3 pods during rollout instead of 4)

Reviewed-on: graham/towerops-web#94

config/runtime.exs

@@ -322,7 +322,7 @@ if config_env() == :prod do
   config :towerops, Towerops.Repo,
     ssl: ssl_config,
     url: database_url,
-    pool_size: String.to_integer(System.get_env("POOL_SIZE") || "15"),
+    pool_size: String.to_integer(System.get_env("POOL_SIZE") || "10"),
     # Proactively ping idle connections every 1s to detect dead SSL connections
     # before they're handed to a real query. Without this, a dead connection
     # sits in the pool until a query checks it out and hangs for the full timeout.

k8s/deployment.yaml

@@ -8,7 +8,7 @@ spec:
   strategy:
     type: RollingUpdate
     rollingUpdate:
-      maxSurge: 100% # Start all new pods at once for fast rollouts
+      maxSurge: 1 # Limit to 3 pods during rollout to avoid exceeding PG max_connections
       maxUnavailable: 0 # Keep all pods running during rollout
   minReadySeconds: 10 # Wait 10s after pod is ready before considering it available
   selector: