Pull Request and Issue templates make generic references to "model"

[nathan-weinberg]

T4R - generally we support users using one of two models in InstructLab - merlinite and granite - but various Pull Request and Issue templates just ask users about "model" inputs and outputs - we should make this clearer as to what users should actually be using for their taxonomy contributions

[wking]

Alternatively, we can ask them to specify which model they used or suggest they copy/paste the output that mentions that. Because today's Granite is presumably going to be a fair bit different from next year's Granite, and the goal with the taxonomy repo is to collect skills/knowledge that are useful somewhere, and not necessarily only skills that are useful for one particular model today. Or is there an expectation that we prune skills/knowledge that are no longer useful for one particular model today? Because I can see maintenance savings in that kind of pruning, but also a chance at making some potential user with a not-all-that-smart base model sad, and balancing the maintenance cost vs. potential use seems hard.

[lehors]

I agree with @wking that we should ask people which model they used (including the version). Getting that info makes it possible to implement various policies later on.