In reconstruction:

In reconstruction:
- Bug fixes in the computation of the MCS effect in the track
  parameter covariances (the main problem was a bad sign in
  cov(x(y),Theta_x(y)) due to the orientation of the spectrometer along z<0).
  It sligtly affects the matching with the trigger (because of the MCS
  in the iron wall) and mainly the covariance matrix extrapolated at vertex (through the front absorber).
- Correct the computation of the covariance matrix of the trigger track
  (also due to the orientation of the spectrometer), which again improve
  the tracker/trigger track matching.
(Philippe P.)