algo

the MR should fix the issue https://redmine.cbm.gsi.de/issues/3149

It contains the following fixes to the CA tracking:

1. bugfix in the Jacobian of the track extrapolation.
   It was introduced during the rewrite of the math
2. tune binning of the hit grid.
   A little decrease of the efficiency happened after the code rewrite
3. tune the cuts for newly introduced hit ranges.
   The new formulae were not tuned
4. revert triplet cut on space + time chi2. 
   After the split of space and time Chi2 in the track fit,
   the chi2 cut was changed w/o tuning. Replace it with the old cut.