use_cross_validation is redundant
I believe the use_cross_validation parameter of the HIPAR constructor is superfluous given the fact that some sklearn classes already include CV in the training procedure. I think it is safest to rely on that existing functionality and allow users to enable CV through the input regression class.