WCOSS2 transition

CORe ran 195001-202401 on Gaea which is a NOAA R&D high performance
computer.  The code was ported to WCOSS2 in order to run it in 
operations.

The WCOSS2 version (202402+) of the code is different from the Gaea version because

   (1) different machine
   (2) newer compiler
   (3) newer libraries
   (4) missing libraries
   (5) NCO code requirements for operations

CORe is built upon gfs v15.  However the gfs v15 build system was not ported to 
WCOSS2.  Instead the gfs v16 build system was available on WCOSS2.  So CORe used 
the gfs v16 build system, and source code from CORe. Changes

  (1) model wouldn't compile, so had to gnu-build of some codes
  (2) some codes wouldn't run, used equivalent gfs v16 versions
        a) using the v16 GSI handles bufr files, and needs to be
               kept current with changes in the bufr format.
        b) cycle of v15 and v16 were similar
  (3) seg fault in humidity table lookup because of humidity stocastic forcing.  
      Fixed but Gaea version would use values outside of the lookup table.  The seg faults 
      were relatively rare and only need to occur in one grid point in large grid of 80 
      ensemble members. So the error was to give the Gaea version a random stocastic humidity
      forcing if a few points.

The CORe code was restructured to run one 6-hour cycle at a time compared with the monthly chunks 
of the Gaea version.  This primarily affected the post processing routines.  The Gaea CORe
ran on 20 nodes of 28 cpus per node.  WCOSS2 (2024) used 128 cpu nodes with faster cpus.  (The Gaea 
nodes that we used were retired 2024-02.)  This allowed 5 ensemble members to run on one
node at one time.  So CORe was able to run in a one node foot print that takes about 3/4 of
an hour for one cycle.