Add label describing if namespace is ower shared runners limits created, pending and running jobs metrics

This MR makes possible to distinguish jobs generated by namespaces that are over shared runners minutes limit from jobs that are truly executed.

As an example - this is the current graph of jobs handled by GitLab.com:

Something about 4000 of pending jobs are jobs created in namespaces that are out of their month limit. These jobs are not send to Runner's so at this moment it's impossible to check how our CI system is behaving. Also our alerting rules, that are based on pending jobs metric, are sending false-positive alerts since two days. Having this change we would get all back to normal and additionally see how many namespaces are out of limit at the moment.