Gen Vertex Miscalculated in L1TrackObjectNtupleMaker

[BenjaminRS]

The current method used to calculate the generator level vertex gives erroneous values. This is particularly apparent in ttbar samples.
Currently the L1TrackObjectNtupleMaker code loops over genParticles collection selecting those with status = 1 (i.e. final state particle) then places the genpartIter->vz() values into a vector. The last entry in the loop is used to assign the gen vertex via the m_pv_MC member variable.
2 Alternative methods to calculate the gen vertex are:

1. Using the HepMCProduct collection and looping through all the vertices. Then looping over all parent particles from a vertex and reject that vertex if parents come from the production vertex. This has been used in other code within L1Trigger/VertexFinder.
2. Using the SimVertex collection (from Geant4 SimHits) which is in the GEN level information. For example, the code has been used in other places in CMSSW.

I have done a preliminary investigation on using the alternative methods and attach the results here. I still need to follow up on an older study by Chris on SimVertex collection to make sure that it is safe everywhere. My preference would be to use the SimVertex collection if it is indeed safe.
GenVertex_230403.pdf