Write tooling to facilitate implementation of metrics collection profiles for OCP components

To support scrape profiles, OpenShift developers would have to deliver as many service/pod monitors as scrape profiles exist (see MON-2693 for more details). The only differences between the monitor resources should be the metricRelabelings/relabelings fields which instruct which metrics to drop/keep. Instead of having developers construct the lists manually, it should be possible to take a "full" monitor resource and provide the equivalent version that keeps only metrics referenced in alerting & recording rules.