Changes for Operations

Changes for operations involved changing the code to meet NCO standards.  
The scientific changes were 

   1.  Bug fixes (seg faults) caused by an out of bounds table lookup
          when humidity reached extreme values.  Think stochastic forcing
          of the humidity/temperature fields.  The lookups were constrained
          to be within the table.

   2.  Some (rare) numerical instabilities caused the model to blow up.
          The time step was changed from 6 dynamical steps per physical
          time step to 8 dynamical steps per physical time step. The
          physical time step was unchanged.

   3.  GSI, enkf and global cycle were changed from gfs v15 to gfs v16.

   4.  CO2 file is not available in real time. So the forecast CO2 was used.

   5.  Prepbufr files will come from the CDAS rather than the CFS dump.
         The CDAS dump has a longer window for data, and so it collects
         more observations.  If the CDAS dump were not used, operations
         would terminate the CDAS dump.

   6.  NCO required changes in the file names.  A perl script is available
         to convert the names to the old format.

There were many compute changes as the system was ported from an older
system that used 20 intel nodes to not-as-old system running on 1 
amd node.  (I wonder how much network time is saved when running the model
on a single node.)