Partitioning the linkml architecture

[hsolbrig]

We have at least three related but distinct artifacts

1. The model (currently meta.yaml + all the includes)
2. The generic tooling built around the model (loaders, dumpers and a portion of the utils directory)
3. The model processing tools (Generators, schemaloader and others)

I would propose that we partition the new linkml organization into

1. linkml-model -- an enhanced instance of the linkml-template, with "meta.yaml" + includes in the source
2. linkml-runtime -- loaders, dumpers and stuff that works with any linkml model
3. linkml -- generators, schema loaders and stuff that works only with the linkml metamodel

We may want to also consider:

1. A generator template
2. separating each generator into a separate repository?
3. separating the mode extensions (extensions, annotations, ...) into (a) separate repository(ies)

[hsolbrig]

We carry an image of the rewrite rules that we use for w3id in tests/test_rewrite_rules. Should we consider a separate repository for this as well?

[hsolbrig]

Chris M - No on separate generator repositories. Also no on extensions (3 repos)