[Mitgcm-support] Merge AIM new-old old-new

[mitgcm-support at dev.mitgcm.org]

Hello,

I will check-in the modified AIM routines:
(no overlap with what Patrick is doing now)

I put a CPP option (in AIM_OPTIONS.h) to reproduce the
same results as Chris version (aimold.5l_LatLon.01).
I checked that the 2 versions give the same results, only when I 
exchange of gUnm1,gVnm1 before solve_for_pressure in Chris version.

Compare to c40-pre9:
* Bug fixed with Katm in aim_do_atmos_physics.F
* diagnostic of surface stress consistent with what is really applied.
* convert surface stress to du/dt : more accurate than it was.
* evaporation is not reset to zero when it becomes < 0 ;

I will start to run the AIM test experiments to update the 
output.txt

Jean-Michel