Changes to angle variables due to possible Awkward 1 bug

@ikrommyd discovered during the validation of the Coffea 2025 version of HiggsDNA, that the angle variable DPhiHJ0J1 contains a bug in line 888 of fiducial.py.

This could be a possible bug of Awkward 1, since that version does not throw an error.

In order for it to work, this line (or its ingredients) has to be changed because of dimension mismatch:

DPhiHJ0J1 = ak.where(HPz == J0J1Pz, -999, DPhiHJ0J1)