Sort out netCDF NC_FILL_FLOAT and NC_FILL_DOUBLE and multiple definitions between 'netcdf_support' and 'grid' modules

[smwesten-usgs]

The 'grid' module actually redefines NC_FILL_FLOAT for use within the module. It is unclear now why the 'netcdf4_support' module was changed from the values defined by Unidata as NC_FILL_FLOAT (9.9692099683868690e+36f) to -9999.0.

[smwesten-usgs]

From netcdf_support.F90 ~line 92:

  real(c_float),  parameter :: NC_FILL_FLOAT   = -9999.0_c_float
  real(c_double), parameter :: NC_FILL_DOUBLE  = -9.9e-20_c_double
!  real(c_float),  parameter :: NC_FILL_FLOAT   = -( HUGE( 0_c_float ) - 1.0_c_float )
!  real(c_double), parameter :: NC_FILL_DOUBLE  = -( HUGE( 0_c_double ) - 1.0_c_double )

From grid.F90 ~line 34:

real(c_float),  parameter :: NC_FILL_FLOAT   = GRID_NODATA_REAL
real(c_double),  parameter :: NC_FILL_DOUBLE   = GRID_NODATA_DOUBLE