PERF(resources): remove CPU limits - keep memory limits only

- CPU throttling prevents app startup, not crashes
- Memory OOM is the real cascading failure cause
- CPU request ensures fair scheduling

kube-state-metrics/helm-values.yaml

@@ -5,8 +5,10 @@ fullnameOverride: kube-state-metrics
 
 resources:
   requests:
-    cpu: 10m
-    memory: 64Mi
+    cpu: 15m
+    memory: 100Mi
+  limits:
+    memory: 150Mi
 
 service:
   type: ClusterIP