Add cf-conventions from technical point of view.

[paolap]

We should include a page explaining why some attributes are particularly important, listing examples of how they are used by software, and what they can break when absent/wrong

[paolap]

This is nearly complete, just need to review maybe add groups??
And possibly implications of multiple time dimensions see #49

[paolap]

I'll try to clarify here the dimensions types and implications and then this should be complete

[paolap]

Reviewing this and this BigData issue:
ACDguide/BigData#88

which implies we should move the content at the bottom of this file to the Governance book:

https://github.com/ACDguide/BigData/blob/main-backup-221027/BigData/format_metadata.md

I think we could add something as:

CDO expect time to be first dimension
xarray can have issues opening files with:

• groups
• several time dimension (see some UM output) what happens is when using open_mfdataset() the variables will be concatenated along more when time axis
• file with conflicting definition of a dimension and it's associated variable i.e:
  dimensions:
  lat = 129 ;
  lon = 185 ;
  variables:
  float lat(lat, lon) ;
  lat:long_name = "latitude" ;
  lat:units = "degrees_north" ;
  lat:standard_name = "latitude" ;
  float lon(lat, lon) ;
  lon:long_name = "longitude" ;
  lon:units = "degrees_east" ;
  lon:standard_name = "longitude" ;

This is definitely bad practice still there are datasets around doing this