Have a per detector retention period

[kaituo]

Is your feature request related to a problem? Please describe.
Currently, we have 90 day retention period for result index. We may want to add per detector retention period.

Describe the solution you'd like
1)We can use our own threadpool to do it, as going through all detector, and all results, is time consuming.
2)If previous delete fails, need to skip current delete.
3)how to deal with failed delete of a detector’s result in current run?
4)Reuse DeleteDetector (code deleted when bumping to 7.8, should find it in commits before) barring the cluster state part
5)Need to compact index since we may have small index after deleting (only those are never gonna be deleted is gonna be saved)