fix(k8s): stop retaining failed cronjob noise

k8s/awoooi-prod/12-cronjob-drift-scanner.yaml

@@ -27,7 +27,10 @@ spec:
   schedule: "0 * * * *"
   concurrencyPolicy: Forbid      # 禁止並發：上次未完成則跳過
   successfulJobsHistoryLimit: 3
-  failedJobsHistoryLimit: 5
+  # 2026-05-24 Codex: failed Job objects make ArgoCD Application health stay
+  # Degraded long after the next scans have recovered. Failure evidence must
+  # live in AwoooP/alert logs, not as retained Kubernetes Job noise.
+  failedJobsHistoryLimit: 0
   startingDeadlineSeconds: 60    # 錯過時間窗口超過 60s 則跳過
   jobTemplate:
     spec:

k8s/awoooi-prod/15-cronjob-km-vectorize.yaml

@@ -22,7 +22,10 @@ spec:
   timeZone: "Asia/Taipei"
   concurrencyPolicy: Forbid
   successfulJobsHistoryLimit: 3
-  failedJobsHistoryLimit: 3
+  # 2026-05-24 Codex: do not retain Failed Jobs in the ArgoCD app tree; stale
+  # failure evidence belongs in AwoooP/KM governance, while retained Job
+  # objects keep the whole Application Degraded after recovery.
+  failedJobsHistoryLimit: 0
   startingDeadlineSeconds: 300
   jobTemplate:
     spec: