Manual cherry-pick of !45526 (merged)
I'm fixing the reweighting problem reported in AGENE-2036 for some processes like DMGG. This was due to the
pdfreweight option is set to 0 in the input card used for reweighting (while it is left to 1 in the input card used for generation, which seems to be harmless). I may try to add the
Another small fix is I'm making sure that the
parallelstage keyword is present for all V2 processes (although it's frozen, so users can't modify it), since it was shown in AGENE-1831 that it could lead to problems.