cnpg: reduce cluster CPU requests to 50m (#247)

Reduce CloudNativePG cluster CPU requests to `50m` across the platform to cut reserved-but-unused CPU.

Updated (memory and limits unchanged):
- artifactapi, authentik, encapi, paperclip (250m → 50m)
- grafana (100m → 50m)
- litellm (250m → 50m)

Excluded: woodpecker (already 50m), puppet (handled separately).
Reviewed-on: #247
Co-authored-by: Ben Vincent <ben@unkin.net>
Co-committed-by: Ben Vincent <ben@unkin.net>

apps/base/artifactapi/cnpg_cluster.yaml

@@ -79,7 +79,7 @@ spec:
       cpu: 500m
       memory: 512Mi
     requests:
-      cpu: 250m
+      cpu: 50m
       memory: 256Mi
   smartShutdownTimeout: 180
   startDelay: 3600

apps/base/authentik/cnpg_cluster.yaml

@@ -79,7 +79,7 @@ spec:
       cpu: 500m
       memory: 512Mi
     requests:
-      cpu: 250m
+      cpu: 50m
       memory: 256Mi
   smartShutdownTimeout: 180
   startDelay: 3600

apps/base/encapi/cnpg_cluster.yaml

@@ -79,7 +79,7 @@ spec:
       cpu: 500m
       memory: 512Mi
     requests:
-      cpu: 250m
+      cpu: 50m
       memory: 256Mi
   smartShutdownTimeout: 180
   startDelay: 3600

apps/base/grafana/cnpg_cluster.yaml

@@ -46,7 +46,7 @@ spec:
       cpu: 500m
       memory: 512Mi
     requests:
-      cpu: 100m
+      cpu: 50m
       memory: 256Mi
   smartShutdownTimeout: 180
   startDelay: 3600

apps/base/litellm/cnpg_cluster.yaml

@@ -79,7 +79,7 @@ spec:
       cpu: "1"
       memory: 1Gi
     requests:
-      cpu: 250m
+      cpu: 50m
       memory: 512Mi
   smartShutdownTimeout: 180
   startDelay: 3600

apps/base/paperclip/cnpg_cluster.yaml

@@ -79,7 +79,7 @@ spec:
       cpu: 500m
       memory: 512Mi
     requests:
-      cpu: 250m
+      cpu: 50m
       memory: 256Mi
   smartShutdownTimeout: 180
   startDelay: 3600