fix: ensure that workloads are updated on logging-operator upgrade

This flag causes e.g. daemonsets and statefulsets to be forcefully
deleted and recreated if required for a configuration change.

Unfortunately this can mean transient data loss (e.g. data in buffers),
but that seems to be a reasonable tradeoff.

charts/lagoon-logging/templates/logging.yaml

@@ -5,6 +5,7 @@ metadata:
   labels:
     {{- include "lagoon-logging.labels" . | nindent 4 }}
 spec:
+  enableRecreateWorkloadOnImmutableFieldChange: true
   fluentd:
     security:
       podSecurityContext: