BDSDipoleStepper doesn't calculate error

[Jochem Snuverink]

BDSDipoleStepper always gives error zero, should be set properly.

[Laurie Nevay]

This is actually correct as it's an analytical solution to the motion in a field. The equivalent Geant4 integrator for a uniform magnetic field also supplies zero error. The dipole integrator is being further developed to ensure robustness for unusual circumstances.

This often seemed to be the cause of various long running tracking bugs, but these have since been found out to be low momenta particles stuck in strong magnetic fields.