Prometheus metrics get overwritten in group execution mode

[deajan]

Push gateway only allows one (the last) metric per job, regardless of labels.

So we have a problem here:

• For textcollector: append to file for the whole run of npbackup allows multiple labels to coexist for same metric
• For push gateway: we need to change backup job to include repo name and action. This will create a new problem, eg persistence

[deajan]

So this actually needs a bit more investigation.
Both textfile collector and push gateway metrics will stall.
Let's keep the partial fix for the push gateway right now.