Serve metrics for non-leader pods #196
Description
When using leader election, metrics aren't served and so the non-leader pods are shown as "Down" by prometheus when it scrapes the pod.
Raised originally in #132 (comment)
To fix this, controller-runtime should start serving metrics while waiting to acquire leader status.
We need to make sure however that it doesn't serve metrics that are false, prometheus should pick up no time series from it and just connect, read and get a basically empty document.