Baseline models - negative binomial

[kathsherratt]

We currently have a few standard comparison/baseline models in the Hub.

A recent evaluation of the 2020 CDC West Nile Virus challenge included a negative binomial model as a standard comparison model, finding it in fact performed very well.

We could think about similarly adding in a simple negative binomial model.

I know there is plenty more thinking to be done around a standard suite of baseline forecasting models in general. I suspect there isn't capacity to act on (any of) this soon, but opening an issue to note the idea.