What should the data N(z) be?

[damonge]

Currently, to get the prior on N(z), we set the mean to the average of the COSMOS-30-band N(z) and the N(z)s from the different photo-z stacks (see here). I think this should just be the COSMOS-30band N(z), since the CV covariance we use corresponds to the COSMOS-30band sample.