LinTzuYaunDataset datasets, Classification problem support added to HGNN

[DanielChaseButterfield]

LinTzuYaunDataset datasets, Classification problem support added to HGNN

Features

• LinTzuYaun Datasets implemented, which use foot positions and velocities with class labels instead of GRF labels.
• Huge Dataset code refactor which allows all datasets to reuse the same get() functions and sorting functions. Even datasets with different inputs can reuse these same functions, as the functions will check the returned dataset arrays and omit any that are None.
• Classification model support added to HGNN models, using Cross-Entropy loss for training and Accuracy for evaluation.
• Mini Cheetah URDF model added.
• Sweeper automatically generates a visualization of the model performance (for regression models).
• A very simple ANN dataset class implemented with limited functionality.

Organization

• datasets.py file seperated into multiple python files in the "datasets_py" folder.

Bug Fixes

• Fixed a bug where the model name saved didn't match the sweeper name.