Fix wrong hit uncertainty in the FT
This fixes an incorrect hit uncertainty of ftHits (on v2::track) which caused some problems in the ParameterizedKalmanFit. Somehow this problem escaped our attention when the Kalman filter was changed. Even this is now fixed, there are still differences between the hit uncertainties used in this version and the ones obtained from the MeasurementProvider. The Pr algorithms use different parametrizations for these uncertainties. E.g.:
Maybe this can be unified at some point.
FYI: @sstahl
NoCutWithFit throughput goes from ~8700 Hz to ~8200 Hz as a result of this fix.
Edited by Marco Cattaneo