Expose config-reloader prometheus metrics

[tommasopozzetti]

Currently, when config-reloader fails to validate the fluentd configurations
for a given namespace, an error is logged and is attached to the namespace
via an annotation. This system, however, requires users to actively look
at namespace annotations or logs to ensure no errors are present whenever
a change in a fluentd configuration is deployed.
Users that are already utilizing Prometheus in their Kubernetes clusters,
are most likely measuring applications' stability via metrics and routing
alerts to a central location, where they are more visible.

This commit introduces the capability for config-reloader to expose its
own Prometheus metrics, which provide a simple boolean metric per
namespace exposing whether the fluentd configs in that namespace passed
validation or not.
These simple metrics allow users to define rules in Prometheus to alert
in case a namespace is throwing errors in fluentd configurations validation.
Such a rule can look something like

alert: FluentdConfigValidationFailure
expr: kube_fluentd_operator_namespace_config_status{} == 0

This commit also adds the possibility of configuring the port that the
config-reloader will be listening on to expose such metrics via a flag.
Furthermore it adds the necessary resources to the helm chart to ensure
that, if the prometheusEnabled value is set to true, the corresponding
Services and, if enabled, ServiceMonitors are also created.

[OrlinVasilev]

@tommasopozzetti great contribution! We will review and test in our labs we will cut a build as soon as we can. Thanks for all the contributions and you effort to this project is appreciated !

[viveksyngh]

LGTM

[viveksyngh]

@tommasopozzetti Thanks for the contribution !!