jra55 tests for different grids
  gx3 uses jra55_gx3
  gx1 uses jra55_gx1
  tx1 is already using jra55 by default (no changes to tests required)
  gbox* uses internal forcing, no jra55 data (no changes to tests required)

remove ocn_data_dir='default' and bgc_data_dir='default' in set_nml.jra55*
files because it causes the bgc tests to fail.  the bgc tests work with WOA bgc
forcing data with jra55 physics forcing data.

iobinary tests must use the old gx3 binary forcing, the jra55_gx3 netcdf data
causes an abort because iobinary is tested with netcdf turned off.  the old
gx3 data is NCAR_bulk in binary format.

alt01 with jra55 uses the alt01a setting.  the difference is that
use_leap_years is set to false in alt01a.  It's set to true by default
in the jra55 datasets.

set_nml.gx3 and set_nml.gx1 are now JRA55 by default
get rid of set_nml.jra55* files
add set_nml.gx3ncarbulk and set_nml.gx1coreii which invokes prior forcing datasets
modify the test suites to do identical tests but with new options, compare
  to prior implementation by creating a renamed regression baseline via
  a script and links to generate the new names.  Use zz* for the new set_nml
  options temporarily just to make it easy to generate the new links
  (the zz ensures the new options are last in the list by alphabet)
chmod -x all set_* files for consistency

this is done as a clean separate step so new baselines can be established to
continue to evolve the test cases and allow regression testing