tendency_of_*_due_to_zhao_carr_physics is still not correct. It should need to recompute precipitation from the fortran state after gscond.