Initial setup, porting/adapting model code from flusion-experiments

[elray1]

This PR does the following:

• Initial package setup with things like the readme and pyproject.toml.
• move modeling code for sarix and gbqr over from flusion-experiments
• adapt that code to be organized around two classes: sarix.SARIXModel and gbqr.GBQRModel.
  • For now, each takes a config object with model settings on instantiation, and has a run method that takes another config object with model run settings.
  • I'm thinking that in this package, it likely makes sense to just have them accept a bunch of named keyword arguments rather than the config objects. To do later.
• Add some integration tests.
  • SARIX model runs are reproducible on different machines, so that test should pass everywhere.
  • GBQR model runs are not reproducible. That test asserts equality of a data frame, and I expect it to pass only on my computer. Automated tests run by github actions fail for that reason.
  • There is an R script in there that can be used to confirm that the expected test data looks plausible; it generates this plot:
    

[elray1]

Noting that tests pass locally on my machine, with a bunch of warnings related to deprecated pandas functionality.

============================================== 3 passed, 21 warnings in 88.84s (0:01:28) ===============================================

[matthewcornell]

approved via live review