ClusterFlow does not react promptly to namespace label changes (real-time reconciliation needed)

[nikozl]

Is your feature request related to a problem? Please describe.

The Logging Operator does not react promptly when a namespace label (e.g., c-logging/enabled: "true") is added or removed. ClusterFlow resources that filter by namespace label only update after a significant delay or after manually restarting the fluentd pods. This causes log collection to be slow or inconsistent for newly labeled namespaces.

Describe the solution you'd like

I would like the Logging Operator to watch for namespace label changes in real time and immediately update ClusterFlow matching. This would ensure that log collection starts or stops promptly when the relevant label is added or removed from a namespace, without requiring manual intervention.

Describe alternatives you've considered

Manually restarting the Fkuentd pods to force reconciliation (not ideal).
Create a custom flow per namespace. This works but the idea is to work only with namespace labels if possible.

Additional context

Logging Operator version: 5.3.0
Example ClusterFlow match configuration:

match:
  - select:
      namespace_labels:
        c-logging/enabled: "true"

The issue affects environments where namespace labels are used to control log collection dynamically.