CRAN release requirements

[pearsonca]

• add CI/CD for testing, linting, check as cran, others?
• finalize API - what's in, what's out
• open capability desires - decide what is in vs out (prefer out), and then ensure API supports future addition without breaking changes (e.g. automatically inferred from data, provided as a mode = ... argument which will add more valid values in the future, etc)
  • support multi-location [both parallelization + information pooling],
  • forecasting other objectives [only support counts currently, want proportions etc])
• any required vignettes/documentation (e.g. is the forecasting intro / decision tree needed?)

[CyGei]

We have a solid MVP of the package at this point. However, I think we are still some way from a CRAN release, and the next important step is to get several people actively using the package and providing feedback.

Once we have feedback from our first users, the goal would be to move the custom model implementations—such as LEMMA, EpiEstim, HHH4, and the Forecasting Hub team models—into separate companion packages, with the model authors taking responsibility for maintaining their respective implementations.

There would still be some work involved in collaborating with those teams and helping them create their R package implementations. But once that structure is in place and we have incorporated early user feedback, I think the core package would be in a good position for a CRAN release.