Dataset for MsProject and LR model retraining

[AKuederle]

This PR attempts to retrain the model of the Ullrich ML classifier on the MsDataset.

Current state:

Dataset

• Missing proper cohort labels
• Need to add loader for clinical info
• Fix loading missing reference info as None instead of empty array (difference between no information provided and 0 Wbs found)

New Model

• Script to validate models on TVS
• Retrained new model on all particpants
• Updated Algo to use new model
• New model only HA
• New model only MS

New model Notes:

• New Model slightly outperforms old Model on entire TVS dataset
• However, performance on explcit example data is worse...
• Most suprising thing is that the new model does not pass the test on the sin-wave dummy data anymore...