Passing prognostic variables as arguments in the subroutines that change them, even in their native module

[aducharne]

During the technical meeting of Sept 11, 2015, we discussed the example of mc in hydrol, which is updated by many subroutines and by ok_freeze, knowing that updates to related diagnostic variables are often needed. It would be very helpful to see where mc is changed, and a good way would be to pass it as argument in the subroutines that change it.

The first thinkg to do is to check that it remains consistent with fortran's rules, and that it doesn't not overload memory.