NUG not consistent or clear in definition of 'coordinate variables'

[ethanrd]

There are two sections in the NUG that define/discuss 'coordinate variables':

1. The "Coordinate Variables" section of the "Components of a NetCDF Data Set" page.
2. The "Coordinate System" section of the "Best Practices" page.

They need to be made consistent and clarified.

Note: This came up in the CF Issue "Updating definition of coordinate variable to account for NUG changes" (#174). It is a long conversation that, at this point, has stalled. Basically, CF in some places says it follows the NUG definition of "coordinate variables" precisely, in others it restricts them to numerical types.