Review _colorbar_definition.json and read callbacks to ensure we use a standard namelist

An ambiguity that has been raised a few times by others new to CSET is 'what variable name do I use for CSET'. If I want to look at winds on pressure levels, is it `x_wind` as seen when I load data in iris, or `zonal_wind_at_pressure_levels`? This leads to trial and error, where jobs fail because they cannot constrain to a single cube, or colorbars are not picked up for specific plots.

This issue captures progress in standardising our set of variable names to use LFRic as the truth (defined in `_stash_to_lfric`). These are reasonably nice as they differentiate the level type in the variable name, such as `zonal_wind_at_pressure_levels`, or `eastward_wind_at_10m`.

Linking #536 as they are similar, except this suggests a practical step forward in reviewing `_colorbar_definition.json` along with read callbacks that we are aligning to LFRic names. This scope could also check diagnostics are using standard names in LFRic (I think age of air diagnostic is not compliant, for example, and still expects 'x_wind').

Out of scope:
- General concepts that map variable names (some kind of field object defined, as in RES). These would be good longer term, but we need something pragmatic ahead of populating RAL4 confs.