Positivity cuts

[comane]

This PR addresses #2058

Main modifications:

• removed load_commondata method from LagrangeSetSpec(DataSetSpec) so that it is directly inherited from DataSetSpec where cuts are loaded (before no cuts were loaded for LagrangeSetSpec type datasets)
• For a LagrangeSetSpec promoted Cuts to InternalCutsWrapper so as to allow for the proper loading of cuts. --> note: rules need now to be passed to DataSetSpec
• renamed kin of XGL and added new process for that positivity dataset
• Changed the names of some INTEG and POS processes

REPORTS

• compare fits: https://vp.nnpdf.science/68xE8NcIQhu2i2hAwhII5A==/Scales0_pdf_report_report.html

[scarlehoff]

Thanks! Only some minor comments. Could you also link in this PR some of the fits you've done using this branch?

[RoyStegeman]

Why this change? The process is DIS_F2C, the only difference is that we use the dataset in a lagrange multiplier. Also, this change is not mirrored in process_options.py

[comane]

Not sure about this one tbh, I think I thought that it might make more sense to have this process_type being under the POS ones.
What are you suggesting?

[scarlehoff]

I think it is better to keep it as a POS_F2C. The positivity are special. I would however call them all POS_F2?
The other reason why I think they should not be called DIS_ is that there's a lot of hidden logic that treats DIS_ processes differently, better to keep the positivity ones differentiated.

Agree that a POS_F2 would be needed in process type (I guess you are already doing it like that but positivity/integrability datasets don't need kinematic coverage so no need to implement one).

[comane]

Just to be sure, would you change all process types starting with POS_F2 with POS_F2 (eg. POS_F2C_CCP -> POS_F2)

What about other processes such as POS_FLL should we just leave them as they are?

[comane]

I did rename all of the POS_F2* process types to POS_F2. I have also added the corresponding process types in process_options.py

[scarlehoff]

Left some small comments, but other than that I think this is good to go.
Make sure you rebase the latests changes in master just to make sure nothing breaks but looks good to me.

[scarlehoff]

I think it is better to keep it as a POS_F2C. The positivity are special. I would however call them all POS_F2?
The other reason why I think they should not be called DIS_ is that there's a lot of hidden logic that treats DIS_ processes differently, better to keep the positivity ones differentiated.

Agree that a POS_F2 would be needed in process type (I guess you are already doing it like that but positivity/integrability datasets don't need kinematic coverage so no need to implement one).