Require minimum livetime after a gap before updating fars
In O3 the pipeline would sometimes come out of a long gap in the middle of a series of bursts. SPIIR would quickly exceed its minimum event threshold, and wind up with unreliable backgrounds. For reasonable far calculations, we need to make sure that we both have enough events, and have waited enough time, before we begin updating fars.
The science issue above is pretty certain, but there's some implementation questions to check:
- Does
livetime
reset after a long gap? - Does
nevent
reset after a long gap? - Do different stats (1w/1d/2h) record and reset
livetime
&nevent
independently?- If so, should they be updated independently for this issue? (i.e. is a long gap for one 2h a long gap for 1w?)
See the relevant changes in spiir-production-Oct27-dev (moved to tdavies__require_livetime_for_fars)