Custom id prefixes

[alex-hh]

Addresses #54 by adding kwargs to run

• Have you added to CHANGELOG.md for this PR?

(btw I had a few dependency issues installing kotsu on my new mac - don't know if worth updating requirements to try to address? I haven't fully resolved how to make pip-compile happy, just had to manually install stuff).

[DBCerigo]

Implementation wise, I think all looks great. The only bit bugging me is the arg names, validation_prefix and model_prefix. I think they are a tad misleading, i.e. they indicate on first reading adding a prefix to the validations IDs, as opposed to setting the name for the validation-esq thing. Getting me on that? I think we could do better. Some ideas:

• changing validation and model everywhere to validator and validatee (I had already been thinking about making the names more general that model etc. given you could the use package outside of data science also, maybe overkill though, probs better names than the one I suggested here also), then arg names could be validator_rename, validatee_rename
• validation_rename, model_rename
• rename_validation, rename_model
• validation_entity_rename, model_entity_rename

(btw I had a few dependency issues installing kotsu on my new mac - don't know if worth updating requirements to try to address? I haven't fully resolved how to make pip-compile happy, just had to manually install stuff).

Yea I just been thinking about that in #53 - maybe we shouldn't be using a compiled ...txt in dev? People just use whatever in dev, and then we always get the latest versions of deps in the CI.

[ali-tny]

the naming thing is super difficult... i think "prefix" and "rename" sound like they're altering the actual name of each model

likewise "validatee" i think is a bit weird

is it a huge problem for sktime that kotsu calls them validators and models? is it a possibility for us to just not name them like that?

an alternative that i find slightly less weird wrt naming is doing something like allowing users to set a entity_type: str attribute or something on the registry itself (that defaults to "validation" or "model"), and then use that name to form the column names?

[DBCerigo]

@alex-hh got any further thoughts on this?

I wouldn't want to move from "validation" to "benchmarks" - I think "benchmarks" is overloaded, as it could be the process (the validation) or the result.

The rest of the ideas I'm open to.