Progpy-inspired base class using Pydantic

[WardLT]

An implementation of a ProgPy-like interface using Pydantic to define the the containers which store state, control, and outputs from a model.

See #2 for further discussion on this topic.

To do list:

• [x] Have the Observer return some diagnostics (e.g., y_err) for each step
• [x] Decide how we should vary control inputs between last and current step. (Deferred, see #4)
• [x] Document how to create a new State class
• [x] Document how to use the Observer
• [x] Give unique names to each parameter within a sequence of parameters
• [x] Moving covariance to a property of the filter Revisiting later
• [x] Adjust terminology
  • Observer to online estimator
  • Control to Inputs
  • Output to Measurement
• [x] Switch back to Py3.9

There are more to-dos in the code, but they would be better-suited for later PRs

[WardLT]

@victorventuri , I'll be ready for your review soon. The points I know I need your input on include:

• [ ] Terminology: Things like "control," "observer," "state," and "state of health" are things we can change later but now is probably a good idea to get them right now
• [ ] What kind of things should I return from the UKF implementation as diagnostics? Right now, it only returns the error between estimated and observed outputs
• [ ] Does the process for defining a new model make sense?
• [ ] What about for defining a new "Observer?" Is there another easy one we should implement to sanity check?

I'll request review when I'm ready. Just writing this down now to give you a heads up and to make sure I don't forget.