Refactor and simplify Job Footprint for running jobs #370
Description
Currently the Job footprint is precomputed with a cron service. This creates burst load, computes a lot of stuff that is probably not needed, and causes delays in when the job footprints are updated. Also currently one has to decide which metrics are part of the job footprint and only these metrics can be filtered and sorted.
A better solution would be to do this on the fly when a job is requested. This would be on the critical path though and therefore needs to be fast. A option is to do the statistics computations in cc-metric-store and have there segment wise pre-computed stats.
A drawback is that for running jobs filtering and sorting cannot be done using SQL queries.
Solution: Only allow sorting for those metrics either only running jobs or completed jobs.