Remove completely kinematics_override

[scarlehoff]

This PR closes #2252, which has required datasets that either have only legacy version or for which it was just not done during the translation.

For HERA data in particular the situation was a bit dire since the plots reported Q2 but the numbers were actually Q. Now it should be Q2 everywhere, which might not be what everybody wants, but at least is correct.
The same is true for the NMC (with a horrible result, but the previous situation was just silly so I've modified the kinematics for that) and NUTEV.

(personally I'd move all the squared quantities to just... not the square?)

Edit: the fecking NMC legacy dataset with kinematics done with a dice had of course to be part of every single regression test 🤷 I hope we can move to the hepdata-binned version.

[github-actions]

Greetings from your nice fit 🤖 !
I have good news for you, I just finished my tasks:

• Fit Name: NNBOT-e9118611b-2025-01-29
• Fit Report wrt master: https://vp.nnpdf.science/a0kifkAWSlSTtrwUa7KqbQ==
• Fit Report wrt latest stable reference: https://vp.nnpdf.science/0unfjZjERuKIlPiwKcWkdA==
• Fit Data: https://data.nnpdf.science/fits/NNBOT-e9118611b-2025-01-29.tar.gz

Check the report carefully, and please buy me a ☕ , or better, a GPU 😉!

[Radonirinaunimi]

Thank you very much for this (especially for fixing the remaining quirks of the old re-implemented sets)!!

So when would the k_ be completely stripped out of the codes? And do you perhaps happen to have a comparison of the kinematic plots?

[scarlehoff]

yes, the k will (in a subsequent PR) all be removed. I cannot do it here because internally validphys uses the f**** kn everywhere so it requires some care.

Regarding the plots, I do have comparisons for the entire set of data. imho when they have changed they look better but I don't want to clutter the server too much, as a surrogate you can look at this comparisons between this comparefit: https://vp.nnpdf.science/a0kifkAWSlSTtrwUa7KqbQ==

and this older one: https://vp.nnpdf.science/eB5jnBi4Qn-OEc2KCq_i1g==

(if you want to look at the whole set of changes I can upload them to the server)

[giacomomagni]

Looks cleaner!!

[Radonirinaunimi]

yes, the k will (in a subsequent PR) all be removed. I cannot do it here because internally validphys uses the f**** kn everywhere so it requires some care.

Okok, good!

Regarding the plots, I do have comparisons for the entire set of data. imho when they have changed they look better but I don't want to clutter the server too much, as a surrogate you can look at this comparisons between this comparefit: https://vp.nnpdf.science/a0kifkAWSlSTtrwUa7KqbQ==

and this older one: https://vp.nnpdf.science/eB5jnBi4Qn-OEc2KCq_i1g==

As far as I can tell, everything looks good.

(if you want to look at the whole set of changes I can upload them to the server)

Maybe even just for the record, this is good to have on the server in any case.

[scarlehoff]

Maybe even just for the record, this is good to have on the server in any case.

I put them in /home/nnpdf2/nnpdf/PR_tests/PR_2266_removal_of_kinematics_override (this nnpdf2 is a second disk we got since the first one was full, I've started moving things around but still lots to do). The folders are separated by <experiment>_master which is run with master and <experiment> which is run with this branch. To first approximation I ran them commit by commit (so the LHCB folder refers to the LHCB commits and so on).