Implement Conway–Maxwell–Poisson likelihood

[ahwillia]

The Poisson distribution is limited to have mean = variance. Stevenson (2016) and Wei & Stevenson(2023) describe a nice alternative that could be incorporated into our GLM module.

There are some subtleties about how to implement this, but I have a prototype here: https://colab.research.google.com/drive/1iFAE1_QUCoFvyxlfyc14G9n4_G0vKRa0?usp=sharing

This would be a nice project for a new contributor.