Reduce Oban concurrency for 3-pod cluster, bump memory limit

Per-pod queues divided by 3 to maintain sane cluster totals.
Lifeline rescue reduced to 10 min to recover OOM orphans faster.
Memory limit 2Gi → 4Gi for HRRR wgrib2 processing.

config/runtime.exs

@@ -117,10 +117,11 @@ if config_env() == :prod do
 
   # Production Oban: live scoring, polling, and on-demand QSO enrichment (no cron backfill)
   config :microwaveprop, Oban,
-    queues: [propagation: 1, commercial: 2, solar: 1, weather: 10, hrrr: 3, terrain: 5, iemre: 10, backfill_enqueue: 1],
+    # Per-pod concurrency (×3 pods = effective cluster total)
+    queues: [propagation: 1, commercial: 1, solar: 1, weather: 3, hrrr: 1, terrain: 2, iemre: 3, backfill_enqueue: 1],
     plugins: [
       {Oban.Plugins.Pruner, max_age: 3600 * 24},
-      {Oban.Plugins.Lifeline, rescue_after: 30 * 60 * 1000},
+      {Oban.Plugins.Lifeline, rescue_after: 10 * 60 * 1000},
       {Oban.Plugins.Cron,
        crontab: [
          {"0 8 * * *", Microwaveprop.Workers.SolarIndexWorker},

k8s/deployment.yaml

@@ -92,7 +92,7 @@ spec:
               memory: 512Mi
             limits:
               cpu: "2"
-              memory: 2Gi
+              memory: 4Gi
           securityContext:
             allowPrivilegeEscalation: false
             runAsNonRoot: true