Consider Kalman Filter Linearization

[teubert]

Requirement Text
Consider automatic Kalman Filter Linearization for non-linear models

Background Information
KF algorithm can only be used with linear models. There are approaches to linearize non-linear models (e.g., DMD surrogate model)

Suggested Solution
Automatically generate linear surrogate model (e.g., DMD) to use in KF if using with non-linear model.

DoD

• Consider if we want to do this.

If yes,

• Implement feature
• Add to example
• Add tests
• Add to change notes for next release

Note
We may not want to do this. The EKF does something similar (see nasa/progpy#30).

Note
Another alternative might be to show an example where someone uses generate_surrogate to create a DMD model and then uses that with the KF

Note
Idea from @matteocorbetta