[CF-metadata] attributes for min/max data values for visualization

[Jonathan Gregory]

Dear Jon

I would prefer the simplest scheme. I am not sure really whether it is a good
idea to have redundant metadata though - we usually try to avoid that. It would
be useful to have more views.

I don't think this would belong in cell_methods, which describes what
statistical reduction has been applied to the data.

It is best to open a trac ticket when there is a precise proposal to be
debated i.e. what changes are to be made in the wording of the conventions
and conformance documents.

Cheers

Jonathan