Fix helm test: drop hook-succeeded so --logs can read the test pod

CI runs `helm test ister -n ister --logs`. The test pod carried a
hook-delete-policy of before-hook-creation,hook-succeeded, so Helm deleted it
the moment it passed — then --logs raced to fetch logs from a pod that no
longer existed and failed with "pods ister-test-connection not found", turning
a green test into a red step.

Drop hook-succeeded; the pod now survives a successful run so its logs are
retrievable, and before-hook-creation still clears it on the next test run.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Author: tijder

templates/tests/test-connection.yaml

@@ -6,7 +6,11 @@ metadata:
     {{- include "ister.labels" (dict "ctx" . "component" "test") | nindent 4 }}
   annotations:
     helm.sh/hook: test
-    helm.sh/hook-delete-policy: before-hook-creation,hook-succeeded
+    # No hook-succeeded: that deletes the pod the instant it passes, and then
+    # `helm test --logs` (how CI runs it) races to fetch logs from a pod that is
+    # already gone — "pods ... not found", a failure on a green test. Keep the pod
+    # after success; before-hook-creation still clears it on the next test run.
+    helm.sh/hook-delete-policy: before-hook-creation
 spec:
   restartPolicy: Never
   automountServiceAccountToken: false