[Mitgcm-support] Crank-Nickelson

[mitgcm-support at dev.mitgcm.org]

Hello,

I have just check-in some modifications:

a) clean up unused CPP_kee in CPP_OPTIONS.h 
   (in model/inc + verification/*/code)
INCLUDE_SHAPIRO_FILTER_CODE
DO_PIPELINED_CORRECTION_STEP     
INCLUDE_FIND_RHO_CALL
INCLUDE_CALC_ISOSLOPES_CALL 
INCLUDE_LAT_CIRC_FFT_FILTER_CODE

and in solve_for_pressure.F 
(remove the old CPP_kee DIVG_IN_DYNAMICS)

b) implementation of the Crank-Nickelson barotropic time 
stepping. It seems to work.
But because I did not modified the 3D solver part, to run in 
non-hydrostatic still requires the use of fully implicit 2D-solver.
(check and stop in ini_params.F)

c) use "zonal_filt_lat.LT.90." as a running flag for
zonal FFT filter: I think its useful since 
FFT filter had an effect (NAN in my case) even with 
zonal_filt_lat=90.  --> modif forward_step.F

Now, the list of modifs since chekcpoint35 is significant, and could
be safe to make a tag now. Alistair, what do you think about this ?

See you,

Jean-Michel