More realistic redshift binning tests

[Lhior]

It seems to me like the current tests done in the testing redshift bins notebook are assuming we have analytical N(z) whereas when we get the data we would rather have the z values for each galaxy and we could obtain the shape of the N(z) by doing a histogram over them.

Should we try to replicate the current tests but doing the binning at the level of doing a histogram over a vector of z values that follow the N(z) distribution?