DiscretePMF type

[dylanhmorris]

Worth considering for interface standardization.

[damonbayer]

Can you say more? What is the relationship between this and the existing DeterministicPMF class?

[dylanhmorris]

The idea would be for DeterministicPMF and any stochastic ones to inherit from it and share attributes (e.g. support for knowing the indexing, knowing the size of the support, etc)

[dylanhmorris]

This was inspired by my realization that we should do this:

https://github.com/CDCgov/pyrenew-hew/blob/52e827dcdac607aa8b75a16e042cf78a02353d0e/pipelines/build_pyrenew_model.py#L124-L133

Which required adding a size() method to this:

https://github.com/CDCgov/pyrenew-hew/blob/52e827dcdac607aa8b75a16e042cf78a02353d0e/pyrenew_hew/pyrenew_hew_model.py#L33

[dylanhmorris]

(Not wedded to the naming conventions)

[dylanhmorris]

@cdc-mitzimorris @damonbayer. I am increasingly convinced that given how fundamental they are to renewal processes, discrete PMFs should be first-class concept in Pyrenew, and that this would make a lot of things (validation!) easier and more DRY. Thoughts?