2d mean continuum: normalized the slope bq per quasar

[julienguy]

There is a degeneracy between the slope of the mean continuum and the mean of the bq terms (slope correction) per quasar.

We can fix that by forcing a mean(bq) = 0 after the fit of aq and bq at each iteration of the delta computation. One has to see if it's possible to fix mean(bq)=0 for each QSO redshift bin independently.

A run with and without this modification and a comparison of the mean continuum in each redshift bin at each iteration would be useful to validate this change.