AthenaCommon+StoreGate: Bypass incidents in AlgContextSvc
As reported/discussed in ATEAM-667 (also else where such as ATLASRECTS-6487), AlgContextSvc
reports a bogus warning about algorithm stack being non-empty at the end of jobs.
This seems to stem from the fact that AlgContextSvc
stops popping algorithms from its internal list on EndEvent
incident. However, since we fire this incident from an algorithm in the AthEndSeq
, that is never cleared up. So, the service thinks (and reports) that there are N
algorithms in the stack at the end of each job, where N
is the number of events.
It seems one can workaround this issue by telling AlgContextSvc
to bypass incidents, which this MR takes advantage of. I'm not 100% why this is not the default and if this would have any negative side effects. Therefore, let me tag some gurus who can chime in @ssnyder, @tsulaia, @leggett, @akraszna, @fwinkl.