More options for time-based reaping

[atombender]

As far as I can see from the code, all dead jobs are deleted. There's also an option to decide what to do with failed jobs.

However, it's a good idea to be able to keep some dead jobs around for historical purposes, regardless of success or failure. In particular, dead job pods can be inspected for data and logs.

For that reason, I'd like an option to keep jobs by age. For example, if we have a backup job that runs every day, I'd like to keep the last 5 days' worth around.

Secondly, keeping failed jobs is okay, but keeping them for ever may not be what we want. A separate age setting for failed jobs would good.

For me, a time-based setting would work. Perhaps specified as command-line flags (--succeeded-job-ttl=<seconds>, --failed-job-ttl=<seconds>) and/or annotations.