Clarify the Conventions for Boundary Variables

[jamesdoyle21]

CF 1.12
Issue #527: Clarify the conventions for boundary variables, especially for auxiliary coordinate variables of more than one dimension, state that there is no default for boundaries, add more information about bounds in section 1.

Ben Comments,
"Need to check if we have a "vertex dimension" defined as CF makes no assumptions about the center of a cell.
"