distinguish no-op when no tasks needed from pipeline failure

See https://github.com/AllenNeuralDynamics/aind-analysis-wrapper-template/issues/18#issuecomment-5082562011

If the analysis wrapper finds no tasks to run (ie all have already been run) it looks like a pipeline failure because no files get written. I'm not convinced this is necessarily a problem, but we should at least have better logging so it looks less like one. Writing a placeholder file might be helpful, but we don't want the placeholder to launch a wrapper capsule so would need to tweak the pipeline connections to avoid that. Also I think it _is_ a problem if a bad query leads to no jobs, so we may want to distinguish that from the case when all jobs are skipped.

There's an analogous question for the wrapper in the case it receives no job records, but that really shouldn't happen so I'm not convinced its worth addressing.