Updating expected TestHepMC minimum efficiency for low Nevents jobs

When jobs generate fewer than 100 events (50 or fewer), then a single ev event can cause them to fall below the required TestHepMC efficiency thr threshold of 98%. This MR updates the default TestHepMC efficiency in su such cases to N/(N+1) to allow jobs that have one failing event to compl complete successfully.

There is some concern that if this happens for many jobs with low numb numbers of events, the effective TestHepMC efficiency will be low enough that it should fail (e.g. if all jobs fail one event, and each job generates 25 events, the total efficiency will be 96% and the jobs should be failing. Perhaps this is something that can be improved in the future by tracking a global TestHepMC efficiency for the tasks via AMI or ProdSys.

Generators/EvgenJobTransforms/share/skel.ABtoEVGEN.py

@@ -325,6 +325,12 @@ else:
     postSeq.CountHepMC.RequestedOutput = evgenConfig.nEventsPerJob if runArgs.maxEvents == -1  else runArgs.maxEvents
     evgenLog.info('Requested output events = '+str(postSeq.CountHepMC.RequestedOutput))
 
+    # Special case of N<100: adjust TestHepMC. We will allow _one_ event to fail the checks.
+    # This means the minimum efficiency is N/N+1 for N generated events. Note that if N<100,
+    # each failed event costs us more than 1% of efficiency.
+    if hasattr(testSeq, "TestHepMC") and postSeq.CountHepMC.RequestedOutput<100:
+        testSeq.TestHepMC.EffFailThreshold = postSeq.CountHepMC.RequestedOutput/(postSeq.CountHepMC.RequestedOutput+1) - 0.01
+
 ## Check that the keywords are in the list of allowed words (and exit if processing an official JO)
 if evgenConfig.keywords:
     from GeneratorConfig.GenConfigHelpers import checkKeywords