Blip improvements

[wforeman]

This PR enables new BlipReco improvements introduced in ubreco (PR 40) so that they can be applied in production.

New information added to the ntuple:

• Predicted blip direction using an RNN, trained on proton blips by Liani
• Flag indicating if the returned RNN direction is valid (RNN requires some minimum criteria which not all blips satisfy)
• Electron lifetime from database: elifetime
• Whether a blip was also a part of a Pandora track, and what fraction of its hits are shared with the track
• More detailed ancestry information on each blip, including
  • the PDG and TrackID of the primary ancestor traced all the way back to the neutrino final-state
  • an integer to help categorize the origin of the blip, specifically with respect to neutron activity

The blip categories encoded in blip_true_category are:
// -9 = no truth match (data/overlay)
// 0 = truth-matched, but not falling in category
// 1 = primary (n,1p)
// 2 = primary (n,Np)
// 3 = secondary (n,1p)
// 4 = secondary (n,Np)
// 5 = primary (n,gamma)
// 6 = secondary (n,gamma)
// 7 = ncapture gamma
// 8 = mu capture gamma

Since these improvements require new capabilities in ubreco/BlipReco, the reconstruction has to be re-run (instead of just reading the blip objects already saved to the art::Event from reco2). A FCL configuration was added to enable the re-running of BlipReco, which is now set to true by default. This is very quick and the added computation time is negligible in comparison to all the other analysis tools that are run in the NeutrinoSelection module.

The advantage to this is that you only need to re-run the ntuple-generation stage on existing reco2 files to get these improvements.