correct DTF (supersedes !610)

Makes major changes to the behavior of DecayTreeFitter:

• MR !610 (merged) introduced a bug where the mass values of the head particle from DTF were often 0. This is due to a problem with the DTF persistency--the dtf object must be preserved in order for the needed values to populate properly.
  • Curiously, this did not pose a problem for the extracted daughter properties.
  • Ideally, we would add checks to the production so that such issues could be spotted more readily in the future, but we've yet to get them to run successfully.
• Instead of running one DTF per \pi^0/\eta with the dielectron PID changed to make DTF treat it as prompt, runs two instances, one with the prompt adjustment, one without. The old dtf branches are gone in favor of pdtf and ddtf for the prompt and displaced cases, respectively.
• Allow particles to be recorded even if DTF doesn't succeed. (DTF values are set to -1 in this case.) Expected to be a small effect.

Instructions

When creating your merge request:

• Add a label for your WG
• Mark as "Draft:" until you're happy with the CI results
• If the merge request supersedes a previous one, add a comment at the end of the header: (supersedes !<MR number>). Please state explicitly in the description if the superseded ntuples can be deleted immediately, or after the new merge request is finalised

When ready:

• Remove the "Draft:" from the title:
• Assign the merge request to you working group's DPA/RTA liaison

To find your working group's DPA/RTA liaison, see this page:

https://twiki.cern.ch/twiki/bin/viewauth/LHCbPhysics/LHCbWGLiaisons