Investigate why residuals are different between py2 and py3

[sguillot]

in tutorial notebooks Modeling.ipynb and Instrument_synergy.ipynb

[thjsal]

Problem solved! It seems that in the Py3 version, GSL seed was still not fixed properly, so we did not get the exact same result. First of all, I instructed you to give gsl_seed input parameter to a wrong function. Secondly, there is also a bug in the seed implementation which prevents fixing the seed to zero from working (and fixing it to zero is what we need here). I'll create a branch to make corrections to the synthesise functions and the tutorials. I already tested that I can get same residuals with Py3 as with Py2, after those corrections.