Potential problem during processing of 4DVAR observation

[arango]

Removed a problem when calling extract_obs3d to interpolate model 3D velocity to observation points. This computation needs the depths a U- and V-points. These depths were computed in a local (allocatable) scratch array. The problem arrised when that array was not allocated. This logic was removed by adding an additional 3D array z_v to mod_grid.F.

Also fixed a bug in make_macros.h during week constraint.