Allow LIMHEL>0 in cudacpp #988
Description
This is a followup to #950.
In 950, which is now complete (and will merged in #955 and #986), @oliviermattelaer did two things amongst others
- LIMHEL in fortran is now controlled by a runcard
- in cudacpp, the runcard sets limhel=0 by default for now (see df130b0)
What remains to be done is to allow limhel to be different from 0 in cudacpp. This can be done with lower priority in a different PR.