Fix for nominal jet pt with postprocessed NanoAOD

Fix for nominal Jet pt with postprocessed, which was random between Jet_pt (from the original NanoAOD, so unsmeared) and Jet_pt_nom (with redone JEC and smeared) because getVarName returned ('pt', 'nom') for both - probably the explanation for the differences observed by @swertz while testing https://gitlab.cern.ch/cp3-cms/bamboo/-/merge_requests/148 . I also added some protections against this happening again (getVarName can always return None to ignore a variation), which broke a test... I'll see if there's an easy fix when cvmfs comes back.

Apologies for the bug and the time it took to investigate.

This will need a rebase of https://gitlab.cern.ch/cp3-cms/bamboo/-/merge_requests/175 (which helped a lot in understanding what was going on here) for that to be useful in testing PRs