Description of problem

Customer is using the External DNS Operator on AWS and on-premise. The External DNS Operator itself provides metrics via `:8080/metrics`. However when an ExternalDNS object is created, the metrics exposed by the ExternalDNS Pod are not exposed and not scraped by Prometheus.

Retrieving metrics via `localhost` works as expected:

sh-5.1$ curl localhost:7979/metrics
# HELP external_dns_controller_last_reconcile_timestamp_seconds Timestamp of last attempted sync with the DNS provider
# TYPE external_dns_controller_last_reconcile_timestamp_seconds gauge
external_dns_controller_last_reconcile_timestamp_seconds 1.7504298579781175e+09
# HELP external_dns_controller_last_sync_timestamp_seconds Timestamp of last successful sync with the DNS provider
# TYPE external_dns_controller_last_sync_timestamp_seconds gauge
external_dns_controller_last_sync_timestamp_seconds 0
# HELP external_dns_controller_no_op_runs_total Number of reconcile loops ending up with no changes on the DNS provider side.
# TYPE external_dns_controller_no_op_runs_total counter
external_dns_controller_no_op_runs_total 0

However the metrics are only exposed for localhost as the ExternalDNS Pod is started with the following arguments:

[..]
      args:
        - '--domain-filter=testextdnsoperator.apacshift.support'
        - '--metrics-address=127.0.0.1:7979'
[..]

Trying to scrape them from outside the Pod fails as the port is only bound locally:

sh-5.1$ curl 10.131.0.26:7979/metrics
curl: (7) Failed to connect to 10.131.0.26 port 7979: Connection refused

The customer expects that the metrics from the ExternalDNS Pod are automatically exposed and automatically scraped (for example via a `PodMonitor` or a `ServiceMonitor`).

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

external-dns-operator.v1.3.0

How reproducible:

Always

Steps to Reproduce:

• Install the External DNS Operator
• Create an ExternalDNS object with any configuration, confirm that the Pod for that object is running
• By default, the "external-dns-operator" namespace is not labeled with "openshift.io/cluster-monitoring: 'true'". Label the namespace with this label so the Platform Prometheus consumes the ServiceMonitors / PodMonitors.
• Check metrics for the namespace where the Operator is installed

Actual results:

No metrics are being scraped, metrics are only exposed on `127.0.0.1`

Expected results:

Metrics are exposed and are being scraped automatically by Prometheus. The customer expects to see "external_dns_*" metrics in Prometheus.

Additional info:

-