Add a separate space for models/rulesets

[vmenger]

Describe the feature
Some pipeline components, the context algorithm most notably, uses a built-in set of rules, that is now packaged in src/clinlp/resources. It would be nice to have a bit more separation between implementation (the context algorithm) and model/rules (the set of trigger terms, in this example) and data (the qualifier cases in the /data dir). For instance, the transformer-based qualifier detectors already have this, as the model is on huggingface, and the implementation in clinlp (and the data being the EMC corpus, separate from both).

This would also be a nice feature for the rule-based entity matcher, when sets of concepts can be released somewhere. But eventually, we might have trained models as well, that need to be shared somewhere.

A use case for the feature
Not really use case, but would improve the structure of clinlp.

Would you like to be involved in development?
Yes, but if anyone has ideas, please comment below

Additional context
N/a