Clarifying how ps is dealt with as auxiliary variable

[martinjuckes]

#391

[martinjuckes]

Schema 1.0 specification of dimensions in the data request does not refer to required auxiliary variables. This is handled in CMOR. This could be added into the schema, or left as a an external piece of information.

[taylor13]

I think I would handle this as an external piece of information because at least some of these auxiliary variables are model dependent. For example, there are certain variables needed to define parametric vertical coordinates (e.g., sigma, or hybrid-sigma), but if you have a different model formulation, you should not include these variables. Some variables might be needed only for certain vertical coordinate models (e.g., we have requested special variables from the Hadley Centre models because they use a height coordinate, rather than a pressure-based atmospheric coordinate. How should we handle these variations on the request?