Incorrect subjet p4 used in PUPPI SD mass correction lookup

[lcorcodilos]

From this HN thread [1], it seems that the PUPPI SD mass correction done in fatJetUncertainties.py is not correct because it uses the subjets with the JECs already applied [2]. While there was a suggestion to do the proper order of corrections externally, I believe it should be changed here as well since there are users who didn't know about this bug (I didn't know about it and I'm sure others are in the same boat).

[1] - https://hypernews.cern.ch/HyperNews/CMS/get/jet-algorithms/562/1/1.html
[2] -

nanoAOD-tools/python/postprocessing/modules/jme/fatJetUncertainties.py

Line 334 in 5827200

groomedP4 = subJets[ jet.subJetIdx1 ].p4() + subJets[ jet.subJetIdx2].p4() #check subjet jecs