[Feature]: Detect and correct step changes in northing calibration

[misi9170]

The northing calibration tools in FLASC currently optimize for a single wind direction bias across the entire history. It would be great to have a method for detecting when there has been a step change in the northing calibration, as can happen when a yaw encoder resets, and a time-dependent northing calibration correction.

Steps could be:

• Detect periods of steady nothing error and step changes in northing error (possibly using a single bias for all time stamps and outlier detection tools, see #36)
• Determine biases for each identified period (using existing tools where possible, possibly by separating periods into distinct dataframes to apply the northing calibration methods)
• Recombine dataframes if necessary

[misi9170]

Need to: average over noise, detect large change (possibly changes in the slope of a cumulative error from a reference?)

Steps:

1. Apply an approach to detect step changes in northing
2. Remove the steps to get a fixed bias
3. Apply exisitng northing calibration algorithm

[aclerc]

This is a feature in wind-up. It uses the ruptures package along with several custom functions to define an optimal northing offset history for each turbine. Happy to discuss further if it's of interest