small improvement in fake track rejection

[pillot]

This contains 2 little modifications of the algorithm that allow to slightly improve the track selection in the end of the tracking:

• enforce the same tracking requirements during track improvement as during the tracking itself (e.g. the default requirement to have at least 3/4 chambers fired in stations 4&5 was soften to 2/4)
• select in priority tracks with more chambers fired (instead of more clusters) when cleaning connected tracks to account for DE overlaps within chambers

Unrelated to that, a small left over in the track data members from the initial porting from AliRoot as been removed (separate commit).

[aphecetche]

@pillot do you want the 3 commits to remain as is or can I squash ?

[pillot]

@aphecetche you can squash.