Linear regression subpackage added to linear dynamical systems

[ncguilbeault]

This PR adds classes and workflows to the Bonsai.ML.LinearDynamicalSystems package to enable the Kalman filter to perform online linear regression. Additionally, several new visualisers were added to the Bonsai.ML.Visualizers package to allow users to visualise multidimensional arrays using OxyPlot's Heatmap.

Linear Regression Subpackage

3 new Bonsai nodes have been added. The KFModelParameters node provides a class which will create the parameters needed to initialise the Kalman filter as a linear regression model. The GridParameters and MultivariatePDF nodes provide methods to modify and access the python model attributes that relate to the creation of a multivariate distribution.

3 new workflows have also been added. They allow for model creation, Bayesian inference, and computing the multivariate pdf.

Linear Dynamical Systems Package

2 new nodes were added for manipulating 2D multidimensional arrays of double. The Slice operator will subsample an array using the specified start/stop indices of the desired columns/rows. The Reshape operator allows multidimensional arrays to be reshaped into a different number of rows/columns.

Visualizers Package

The HeatMapSeriesOxyPlotBase class provides underlying access to the HeatMapSeries from the OxyPlot package. Both the MultivariatePDFVisualizer and the MultidimensionalArrayVisualizer utilise this base class for visualising data.

[glopesdev]

Looks really nice overall, just went through everything and found a few inconsistencies on documentation, formatting and possibly naming and design conventions. Otherwise all good!

[ncguilbeault]

Thanks a lot for your feedback @glopesdev. I've implemented essentially all of your suggestions and pushed them as new commits.