fix to !512

Issues arising from "rarer" decays (wrt D2hh,D2hhh,D2KSh) in the previous AnaProd- !512 (merged) (submitted but not yet running):

• in some ntuple the decay descriptors didn't use the intermediate combinations used in Hlt2 (eg. J/psi -> mu mu, K* -> K- pi+, ...) causing empty ntuples despite the existing candidates after the filter
• in others the code resulted to be bugged as crushing when running due to typos
• the tuples for D0->KSKS decays have also been fixed

All of this is fixed in this MR. The chain Moore+DaVinci has already been tested locally running over 3M events.