fix(ci): retain completed pods 10 min for log access (CPU only)

Add deleteDelayDuration: 600s to podGC on CPU workflow templates and
sensors (ci-pipeline, compile-deploy, build-image). GPU training
workflows keep immediate cleanup to avoid wasting expensive GPU time.

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

infra/k8s/argo/ci-pipeline-template.yaml

@@ -17,6 +17,7 @@ spec:
       app.kubernetes.io/part-of: foxhunt
   podGC:
     strategy: OnPodCompletion
+    deleteDelayDuration: 600s
   ttlStrategy:
     secondsAfterCompletion: 3600
   volumes:

infra/k8s/argo/compile-and-deploy-template.yaml

@@ -24,6 +24,7 @@ spec:
       app.kubernetes.io/component: compile-and-deploy
   podGC:
     strategy: OnPodCompletion
+    deleteDelayDuration: 600s
   ttlStrategy:
     secondsAfterCompletion: 3600
   activeDeadlineSeconds: 3600  # 1 hour (compile + deploy)

infra/k8s/argo/events/ci-pipeline-sensor.yaml

@@ -46,6 +46,7 @@ spec:
                 serviceAccountName: argo-workflow
                 podGC:
                   strategy: OnPodCompletion
+                  deleteDelayDuration: 600s
                 ttlStrategy:
                   secondsAfterCompletion: 3600
                 workflowTemplateRef:

infra/k8s/argo/events/workflow-trigger-sensor.yaml

@@ -63,6 +63,7 @@ spec:
                 serviceAccountName: argo-workflow
                 podGC:
                   strategy: OnPodCompletion
+                  deleteDelayDuration: 600s
                 ttlStrategy:
                   secondsAfterCompletion: 3600
                 workflowTemplateRef:
@@ -193,6 +194,7 @@ spec:
                 serviceAccountName: argo-workflow
                 podGC:
                   strategy: OnPodCompletion
+                  deleteDelayDuration: 600s
                 ttlStrategy:
                   secondsAfterCompletion: 3600
                 workflowTemplateRef: