Refactor models into different files

[jasonmcewen]

At present all models are implemented in model.py. This file is currently also ~2000 lines long. Let's factor out the different models into separate files, keeping the abstract class in, say, model.py. Then we can more easily add new models.

[jasonmcewen]

Now we have flow support, we should put the abstract class in one file, then the legacy models in another, and the flow models in another.

[jasonmcewen]

We discussed this. The plan is to have the abstract class in model_abstract.py and then to have model.py for the flows and model_legacy.pyx for the legacy ML methods.