vertices_latitude / vertices_longitude should not have attribute units

[larsbuntemeyer]

I am not sure why this is the case for the CMIP6_grids.json table but i think we should not have the units attribute in the boundary variables according to CF conventions:

cordex-cmip6-cmor-tables/Tables/CORDEX-CMIP6_grids.json

Lines 137 to 156 in 80f7f51

	"vertices_latitude": {
	"standard_name": "",
	"units": "degrees_north",
	"long_name": "",
	"dimensions": "vertices longitude latitude",
	"out_name": "vertices_latitude",
	"valid_min": "-90.0",
	"valid_max": "90.0",
	"type": "double"
	},
	"vertices_longitude": {
	"standard_name": "",
	"units": "degrees_east",
	"long_name": "",
	"dimensions": "vertices longitude latitude",
	"out_name": "vertices_longitude",
	"valid_min": "-180.0",
	"valid_max": "180.0",
	"type": "double"
	}

[larsbuntemeyer]

see PCMDI/cmor#729

[gnikulin]

I agree, the units attribute in the boundary variables is redundant information and the boundary variables shouldn't have any attributes, also consistent with the CF conventions.

[larsbuntemeyer]

OK, we can't really do anything about what cmor will write out here i guess. But the way it is now should be fine since these are recommendations. I'll leave the issue open anyway..