Graph model classes

[bdpedigo]

Reference Issues/PRs

Closes #142
Closes #143
Closes #174

What does this implement/fix? Explain your changes.

Adds classes for fitting unweighted graph models to data, including ER, DCER, SBM, DCSBM, and RDPG

Any other comments?

• Does not support weights
• There are many ways to estimate communities in the a posteriori SBM and DCSBM. Here, only one is implemented, but we could support more or find which methods are best in the future
• There are several different ways to model the relationship between degree corrections and community membership in the DCSBM. For now, the degree correction parameters are tied to a simple block, but another model would be to randomly sample a community membership and independently sample a degree correction parameter

[bdpedigo]

@j1c @hhelm10 ready for review

[hhelm10]

looks good on first pass, most comments are for naming "issues"

[j1c]

FYI, this PR also has deleted notebooks. Not sure if that was intentional, but let's be a bit more careful what we add/push to which branch.

[bdpedigo]

Yeah I realized that, I can add notebooks back to this PR but I didn't really see the point in that now

[bdpedigo]

@j1c jovo wants me to just merge this and fix any issues later just cause we need to submit graspy to jmlr. I will probably do that Monday

[bdpedigo]

still open to discussion/improvements on what was merged here since it didn't get a real review