Issue15 matthew mizielinski CVs and tables fixes

[matthew-mizielinski]

Changes I've made to get this past CMOR:

• Inserted "Table" prefix into table_id in every MIP table
• Changed mip_era from "CMIP6" to "SNAPSI" everywhere (CMOR will look for tables & CVs named "CMIP6_*" otherwise)
• Experiment definitions,
  • The activity_id and parent_experiment_id fields need to be lists
  • I've added a list of allowed_model_components with a default of AOGCM and AER. The required_model_components can also be used if there are requirements on experiments.
• SNAPSI_coordinate.json was invalid as the entry for the plev137 pressure level set was not closed properly. Note that this entry may be missing a number of required fields.
• coordinate set was also missing sdepth1 dimension which is used in mrsos in the 6hrPt table.
• Set Conventions field in the grids definition table to be consistent with else where.

Other changes:

• Updated data_specs_version to 0.9.3 and ensured that this format is allowed by the CVs
• Updated table_date consistently to today's date
• Simplified license pattern -- have a careful look at the license you are using as this pattern is painful to get correct. CMIP6 caused itself some issues by using the ShareAlike license.
• tracking_id pattern in CVs set to SNAPSI/.* -- CMOR doesn't like absent prefixes and it needs to include the .* pattern.

A number of the above should be reviewed and possibly altered depending on your purposes. In particular the use of allowed_model_components / required_model_components in the experiment definitions, the plev137 pressure level entry and the license pattern should be considered further.

P.S. The edits to the CVs were made a little crudely. Might be worth pushing through the JSON library to get a cleaner structure (python -m json.tool <input file> would clean this up).

[matthew-mizielinski]

Note that @charliepascoe has a different take on the table_id field (see #12) -- I've had to restore the Table prefix in this pull request to ensure that our code around (but not in) CMOR works as this is the structure we are used to with the CMIP6 tables.

[charliepascoe]

Thanks Matthew, I now need to remember to pull this update into the MIP tables that ceda-cc references.

[matthew-mizielinski]

@charliepascoe, you might be able to shortcut that process using the git submodule command -- I've seen this done in other repositories and then you may not have to manually copy files across.