Feat: Adding relevant fields to save the rich SPINE output

[pgranger23]

Initial proposal on how to include relevant output fields from SPINE into the CAFs.
This is a first guess to start the discussion, I'm open to doing things differently.
Note that many of those fields are genuinely useful and could be populated by other reconstruction algorithms.

[pgranger23]

Following the comments from our meeting:

Summary of Changes since initial submission

1. Rename DLP to SPINE

• Renamed all occurrences of dlp/DLP to spine/SPINE in the Standard Record branch names, enums, files, and classes (e.g., rec.common.ixn.spine, kNDLArSPINETrack, etc.).

2. Simplify Reconstructed Particles collection (SRInteraction::part)

• Eliminated the wrapper class SRRecoParticlesBranch completely.
• Changed part inside SRInteraction to be a direct std::vector<SRRecoParticle>. This avoids redundant parallel lists per reconstruction stack (since the parent interaction is already split by reco stack at the top-level).
• Simplified navigation lookups (Navigate.ixx) and interaction containment logic to loop over part directly.

3. Merge and Rename Detector-Specific Particle Metrics (SRLArRecoParticle)

• Removed cathode-crossing, drift, and clustering metrics from the generic SRRecoParticle (which lives in common).
• Renamed SRPFP to SRLArRecoParticle to represent a general, unified detector-specific reconstructed LAr particle.
• Merged voxel-clustering/fragments metrics (formerly SRClusteringInfo) and SPINE-specific metrics (formerly SRSpineInfo / SRLArTPCInfo), including pid and shape classifications, into the new SRLArRecoParticle class.
• Renamed the Far Detector collection inside SRFDInt from pfps to lar_particles (e.g. rec.fd.hd.pandora.lar_particles).

4. Modularize PID & Energy Hypotheses (SRPIDProbability)

• Grouped kinetic energy estimators (calorimetric, CSDA range, and MCS deflection) under a new SREnergy class nested within SRPIDProbability.
• Species-specific parameters are stored as a std::vector<SRPIDHypothesis> hyp; where each hypothesis contains the PDG ID, score, fit $\chi^2$, and the corresponding SREnergy estimates.
• Implemented map-like helper methods on SRPIDProbability for analyzer-facing convenience:
  • float PIDScore(int pdg) const
  • float Chi2(int pdg) const
  • const SREnergy& Energy(int pdg) const (returns a static default with NaN if not found)
• Measured, hypothesis-independent variables (depositions, start_dedx) remain directly in SRRecoParticle.

[chenel]

I think we're more or less agreed on the fundamentals now, just waiting for signoff from ND.

One note: I think you missed an instance of DLP (probably because it's got the full name in it): line 205 of SREnums.h. Let's change kDeepLearnPhys to kSPINE. Also affects its usage (SRNDLAr.cxx etc.). All the affected spots, I think:

https://github.com/pgranger23/duneanaobj/blob/feat/dlp-detailed/duneanaobj/StandardRecord/SREnums.h#L205
https://github.com/pgranger23/duneanaobj/blob/feat/dlp-detailed/duneanaobj/StandardRecord/SRNDLAr.cxx#L12
https://github.com/pgranger23/duneanaobj/blob/feat/dlp-detailed/duneanaobj/StandardRecord/SRNeutrinoHypothesisBranch.h#L19

[pgranger23]

Good catch, had missed it, thanks!
Should be fixed by now.