Description of problem:

The telemetry instrumentation of Cluster Logging Operator produces stale metrics containing previous and intermediate configuration. For example, creating a ClusterLogging configuration referencing the default LokiStack log storage produces an "empty" set of metrics (all options set to zero) and the "current" set of metrics:

# Example for log_forwarder_output_info metric:
log_forwarder_output_info{azureMonitor="0", cloudwatch="0", default="0", elasticsearch="0", fluentdForward="0", googleCloudLogging="0", http="0", kafka="0", loki="0", splunk="0", syslog="0"} 0
log_forwarder_output_info{azureMonitor="0", cloudwatch="0", default="0", elasticsearch="0", fluentdForward="0", googleCloudLogging="0", http="0", kafka="0", loki="1", splunk="0", syslog="0"} 1

Similarly, switching from ElasticSearch to Loki leads to the old ElasticSearch metrics still being present in the output (with a value of 0).

Version-Release number of selected component (if applicable):

5.9.0

Steps to Reproduce:

1. Install Cluster Logging Operator
2. Configure a ClusterLogging resource
3. Look at the metrics from the operator (prefixed with "log_")

Actual results:

Metric set contains stale metrics.

Expected results:

Metrics should reflect the current state of the configuration only.