#17 Wrong errors and p values when fitting Helium ions

[Anonymous]

When a sample of 1 GeV He2+ ions are simuilated in the basf2 framework and fitted with genfit the p values when using the CDC are almost all 0.

When using the VXD there is a large difference in the p values form the forward and backward filter.

Inspecting the the pull distribution of the genfit parameters individually for each layer revealed that in the forward filter the errors for q/p are way too small and in the backward filter they are way too large.

This indicates a problem in the noiseBetheBloch function

[Tobias Schlüter]

Note that the error propagation from dedx to q/p missed the factor charge^2, which I fixed in my recent commit even though I forgot to mention this in the commit message. Can you check this again?