N3LO IHOU are correlated

[felixhekhorn]

n3lo_cf_variation seems to be a single number, although I believe it should be 4 numbers: (C2g, C2q, CLg, CLq), i.e. one per coefficient we want to vary - similar to the case in eko