improve analysis of requed jobs

[mathisloevenich]

At the moment if a job gets requed only the progress, times, resources, etc. is captured
of the last node it was executed on.

This means the information about previous processes is lost.
However it's captured, that a job was requeued earlier.

At the moment we don't have checkpointing, which means that the information is not necessarily useful.
However it messes up the waiting time as the submission event only happens once, but the last execution event is taken to calculate the waiting time.

[mathisloevenich]

To solve this, allow multiple execution events to be a list to extract various time differences (also per node).