NC_HAS_BYTERANGE is incorrect in netcdf_meta.h in both autotools and cmake build

[edwardhartnett]

Here's a netcdf_meta.h generated by today's master with autotools. Note the value of NC_HAS_BYTERANGE.

#define NC_HAS_NC2       1 /*!< API version 2 support. */
#define NC_HAS_NC4       1 /*!< API version 4 support. */
#define NC_HAS_HDF4      1 /*!< HDF4 support. */
#define NC_HAS_HDF5      1 /*!< HDF5 support. */
#define NC_HAS_SZIP      1 /*!< szip support (HDF5 only) */
#define NC_HAS_SZIP_WRITE 1 /*!< szip write support (HDF5 only) */
#define NC_HAS_DAP2      1 /*!< DAP2 support. */
#define NC_HAS_DAP4      1 /*!< DAP4 support. */
#define NC_HAS_BYTERANGE no
#define NC_HAS_DISKLESS  1 /*!< diskless support. */
#define NC_HAS_MMAP      0 /*!< mmap support. */
#define NC_HAS_JNA       0 /*!< jna support. */
#define NC_HAS_PNETCDF   0 /*!< PnetCDF support. */
#define NC_HAS_PARALLEL4 0 /*!< parallel IO support via HDF5 */
#define NC_HAS_PARALLEL  0 /*!< parallel IO support via HDF5 and/or PnetCDF. */

The problem is in netcdf_meta.h.in, which has:

#define NC_HAS_DAP4      @NC_HAS_DAP4@ /*!< DAP4 support. */
#define NC_HAS_BYTERANGE @HAS_BYTERANGE@
#define NC_HAS_DISKLESS  @NC_HAS_DISKLESS@ /*!< diskless support. */

Like the lines that surround it, byterange needs to use the NC_HAS_ macro, not the HAS_ macro, to get 0/1 instead of no/yes.

I have a fix I will put up shortly.