Covariance matrix in ReconstructedParticle is misleading

[tmadlener]

Looking at #422 I realized that there is an inconsistency or at least potential for confusion in the reconstructed particle. On the one hand we have a momentum, as well as a mass and an energy, where we explicitly state for the latter two that the Four momentum state is not kept consistent internally.

How do we set the covariance matrix in this case? Do we use the mass? Do we use the energy? I think that if we don't keep the momentum, mass / energy consistent, how can we provide a covariance matrix, apart from making it effectively a block matrix (separating momentum covariance and mass resolution). But then we don't need the 4D covariance matrix any longer.

[m-fila]

Thanks! I just had the same though, but I assumed that for sure I'm missing something 😉