Optimize track clust assoc

The track-cluster association algorithm is the most CPU consuming task of TCC and UFO building. The main reason is it uses track extrapolation in the calorimeter. However in the TCC building steps, only the track extrapolation to the entry of the calorimeter is used.

This MR change the association algorithm so the extrapolation is not done through the calorimeter, It also adds an option to only associated the tracks from PV0 which is useful if only UFO is run since UFO only uses PV0 tracks. Finally it also optimizes some other parts of the association alg.

All in all there's a x3 to x15 (when only PV0 tracks) CPU improvement.

All tests have shown identical physics results in the TCC and UFO particles.

I keep this in WIP just to leave time to TCC authors to comment (@sschramm, @rjansky, @ncalace )