Review interdataset correlations

[scarlehoff]

During the implementation of the new data sometimes the uncertainties have changed name, and, when done consistently within a PR, it might have escaped testing.

This should be revised more generally, but let's keep track of the ones we know for now:

• [x] ATLAS luminosity uncertainity for the 2012 data: list The name should be ATLASLUMI12, it should be updated in the top data link @t7phy

Note: keeping the old names for the uncertainties when they are correlated between dataset will allow us to mix legacy and not legacy datasets, which sooner or later will be necessary.