rationalise the use of IDs & Version w/in CIM instances

[allyn]

Scattered throughout the CIM are identifiers and versions referring to various artefacts: the document, schema itself, the controlled vocabularies used by the document, the thingy being described by the document (the actual model or experiment or dataset etc.).

Some of these are strings, others are integers, and others are GUIDs. Some of these are mandatory, others are optional.

This ticket aims to rationalise the way that these IDs and versions are used within the CIM. This will require some structural changes to the CIM. But it will also require some documentation describing when a change to a CIM document warrants a new instance complete with a new id, and when just the same instance with an old version will suffice (and when the whole thing should get overwritten)?

[allyn]

Whatever is decided, a [top-level] CIM document should be citable.

[allyn]

Also, an overzealous use of GUIDs is to be discouraged; computer programs like to name things with random strings of letters and numbers, people like to give things meaningful names. So requiring a model to have a GUID instead of something like "HadGEM3" is a bad idea.