TRestTrackReconnection not properly identifying end-tracks

[jgalan]

This issue is created to follow up a problem discussed on the following forum post

http://rest-forum.unizar.es/t/pandaxiii-mc-pipeline-at-development-v2-3/410/15

The problem is that normally TRestTrackReconnectionProcess should shift the order of hits so that the longer edge is suppressed (minimising the total track length). In other words, just re-ordering in the array so that hit_0 and hit_N correspond with the long edge.

Also the reconnection routine, such as breaking tracks and find an optimum reconnection should solve the problem observed on the track observed at the forum post. This last, fixing reconnection through sub-tracks, might just need optimisation of fNSigmas parameter.