EpiModel inner constructor for ingesting distributions directly.

[SamuelBrand1]

As mentioned here. A step in getting the toy model functional is to improve ability to ingest distributions and operate on their discretisations.

[zsusswein]

A bunch of really nice stuff here Sam. I left some nitpicky comments inline, but nothing major.

My only high-level question is how do we pass the latent process through to the EpiModel struct? Both pieces are in here, but I don't see where they'll end up fitting together.

[zsusswein]

All looks good.

Abstracting the sparse matrix constructor into a new function and if/how we check for GI normalization inline are still outstanding. I don't mind if that's addressed here or in a future issue though.

[seabbs]

I'm happy to do another pass on this tomorrow but for now flagging that we should either make unresolved conversations into issues or address here and rerevie.

[SamuelBrand1]

@seabbs I'm treating the comment about a utility for the sparse delay kernel as an issue.

[zsusswein]

@seabbs I'm treating the comment about a utility for the sparse delay kernel as an issue.

Sounds good. Can you link the issue in the comment thread so that it's easy to track the thread back?

[seabbs]

and then resolve the conversation (as unresolveed conversations are PR blockers)

[seabbs]

Mostly looks good but a few potential merge issues and one clarifying question