Are all samples affected or could we do a partial production? If the former we might as well fic AK8 JECs before starting

Only samples associated to the three modified schemes are affected, but we are talking about many. I would fix any pending issue before proceeding with resubmission.

I agree with Bruno.
Let's fix also the AK8 stuff before resubmitting.
And thanks for all your work :)

The latest commit fixes the issue reported here, which highlights a bug in the LHE weights of the ST_s-channel_4f_leptonDecays_TuneCP5_13TeV-amcatnlo-pythia8 sample. The fix consists on multiplying all weights by two (2.), except the ones with indices 0 and 9.

Added parser to endRun to ensure the PDF ID of the PDF used as reference for QCD scale variations matches the PDF ID of the PDF used as reference for the PDF variations, as required by the GEN contact.
I would have ideally added the logic to beginRun, but that seems not to be supported by default, due to the following error message:

::getByToken: An attempt was made to read a Run product before endRun() was called.

The current solution is however acceptable, I believe.
I'm currently running tests on the single top samples.

The latest fix addresses this issue, where $\alpha_{\text{S}}$ variations are set to zero for the schemes when they are not available. I was previously considering a different PDF set (different PDF ID), which is not the recommended approach. Now the reference PDF for the QCD scale variations should always match the reference for the PDF variations (which makes intuitive sense); this is ensured by the logic in HTauTauNtuplizer::endRun(...).

I have tested locally a single sample for each scheme, but I cannot guarantee there is no corner case where this code fails.

Please review.