Draft: moving to build 4-vec with px, py, pz, m rather than px, py, pz, e

From discussion with @dgillber and @abuckley, move to build TruthParticle 4-vec from px, py, pz, m rather than px, py, pz, e, except for the case when the mass (or energy?) is negative. In this case, one continues to build them with px, py, pz, e. Note that LHE/HepMC provides px, py, pz, e and m values.

This provides "reasonable" values for the mass from the 4-vectors, in particular for electrons which are in general far off 0.511 MeV when using e. I verified with H4l minitree comparison with and without the changes and only masses appear different above 0.001 in fractional difference.