Update resources for tests, examples and documentation to be NNPDF4.0 based

[scarlehoff]

The main point of the changes is for the runcards to be NNPDF4.0 runcards (and not "legacy" runcard that are only supported via hacks) but of course a few other thing change (like the fact that some exponents are now limited by the integrability, for instance).

The rule of the thumb for the changes is:

• Use NNPDF release PDFs when a PDF is needed.
• Change 3.1 NLO theory to 4.0 NLO theory (and 3.1 to 4.0 NLO fit)
• Use a fit to theory 162 (with 4.0 methodology) when a NNLO fit is needed.

TODO:

• Change the n3fit regression tests
• Change the validphys examples
• Change the n3fit examples
• Change the documentation to use the example resources

Some vp runcards need to be modified beyond just the resources being used for them to work:

• generate_a_report.yaml: expects experiment key which doesn't exist anymore
• future_test_example.yaml: actually, maybe it is too complicated for an example and should go the reproduce NNPDF40 folder
• spiderplot_dataspecs.yaml: uses the 1000 replica fit
• spiderplots.yaml: 1000 replica fit

Closes #1432.
Closes #1448.

[scarlehoff]

For the n3fit examples I'm going to still use theory 200 at times since I would like to have integrability datasets in the examples. However, would it be possible to update theory 162 with just one integrability dataset? (copying an arbitrary one from theory 200 would be enough) /cc @enocera

Otherwise I leave theory 200 for the examples which I think is also ok.

[enocera]

@scarlehoff OK, no problem. Do you have any preference for the integrability data set?

[scarlehoff]

Not really. It's just so that's available in the examples (and tested I guess).

[scarlehoff]

Probably needs a very close review to make sure the changes are consistent. I've been trying to test many of the runcards in the examples but there are many

[Zaharid]

I'd like to merge this and then release a new version.

[enocera]

For the n3fit examples I'm going to still use theory 200 at times since I would like to have integrability datasets in the examples. However, would it be possible to update theory 162 with just one integrability dataset? (copying an arbitrary one from theory 200 would be enough) /cc @enocera

Otherwise I leave theory 200 for the examples which I think is also ok.

@scarlehoff I've just realised that I've never mentioned that integrability FK tables have been computed for theory 162.

[scarlehoff]

@scarlehoff I've just realised that I've never mentioned that integrability FK tables have been computed for theory 162.

I'll update the regression test to include integrability then. Thanks!

I'd like to merge this and then release a new version.

Ok, let me change first the name of the low precision fit and add integrability set to the "quickcards"

[scarlehoff]

I just realised the closure test fit was also done with theory 53 so I need to redo that one as well, sorry.

[scarlehoff]

This should be done now.