Apr 27 - May 08, 2012
- update prm code to work with latest MITgcm code (checkpoint63m)
- nullify pointer fData2d before using it (needed with ifort F90 compiler
  recent version, e.g., v 12.0)
- change sourced files (do_links, do_genmake2, do_make, do_run, do_clean ...)
  to normal bash script; update README.

checkpoint63m (2012/05/09)

May 29 : add dir content description in multi_comp_setup/doc/contents.txt file
Jun 04 :
- add this file
- do_genmake2: allow to pass 1 argument to genmake2 command (e.g., -ieee or -devel)
- modify fg code to enable 1 fg model to handle multiple 2-D slices as multiple
  disconnected tiles: requires to #undef SINGLE_TILE in computeFG/computefg.F90
  and to define DISCONNECTED_TILES in fg/code/CPP_EEOPTIONS.h

checkpoint63n (2012/06/04)

Jun 04 :
- use only 1 FG per proc with 100 disconnected tiles (instead of
  100 single-tile FG) but keep the same results (when compiled with -ieee):
   change SINGLE_TILE to #undef in computeFG/computefg.F90
   change DISCONNECTED_TILES to #define in fg/code/CPP_EEOPTIONS.h
  and update loaded input files (inputs/fg/data)
Jun 05 :
 - put addition of initial noise to FGs in fg/code/ini_theta.F (instead of in
   computeFG/computefg.F90): this fixes exchange issue + get valid initial temp
   output ; results change but stay identical when using 100 single-tile FGs.
 - reduce ovelap size from 3 to 2 in both CG and FGs (results unchanged).