add Down category, Never consider Down-Down DiElectron output, filter always...

From K*gamma , on reconstructed candidates, it seems like the Long-Downstream pair is needed to be more efficient.

To make benefit of DEDX cut, the lines are now extended to include Long-Downstream pairs, but veto out Down-Down. To keep rate controlled , all Down-Down from DiElectronMaker are rejected. I think for soft photons or hard photons have whatever the Upstream tracks gives should be cut at PT >0 , while di-electron cuts at 500. I think this is reasonable to have.

Long-Long
Long-Upstream
Upstream-Upstream 
Long-Downstream
Upstream-Downstream

All Long and Upstream tracks include the MEDIANDEDX cut.

@clangenb in attachment the 'reconstructed' tracks categories in K*gamma with gamma converting before z = 2.4 meters. X axis the e+ and Y axis the e-

0 = Not Reconstructed 1 = Long 2 = Down 3 = Upstream