Momentum Parameterisation in Forward tracking not respecting B field symmetries

In the code at https://gitlab.cern.ch/lhcb/Allen/-/blob/master/device/SciFi/looking_forward/include/LFMomentumEstimation.cuh?ref_type=heads#L65 it seems like the final implementation of the parameterisation included terms which are breaking the field symmetries effect on slopes power counting.

There are terms in t_{y}^{5} while all othe others defining C0,1,2 have t_y at even powers. In principle the parameterisation was originally determined using polynomials respecting magnetic field symmetries, so i think this has to be fixed. Or better, move the forward parameterisation for q/p determination to be the same as matching.