Start refactor with class skeleton

[AuguB]

Rough outline of the responsibilities:

NormData

• Loading data from different formats (csv, nifti, cifti, fsl output)
• Checking validity and safety of data
• Holding train, test, val data
• Holding predictions
• Ensuring shape consistency
• Provide convenient slicing and indexing

NormConf

• Hold cross-validation configuration
• Hold logging configuration (output directory, and logging folder and format)

NormBase

• Contains the logic for cross-validation
• Has concrete, model-agnostic implementations of fit, predict, fit-predict methods which receive and return NormData objects
• Calls the abstract fit_model, predict_model, fit_predict_model which receive and return NormData objects
• Performs basis expansions

NormHBR (example concrete implementation)

• Has concrete implementations of fit_model, predict_model, and fit_predict_model, which receive and return NormData objects
• Acts as an adapter to the HBR model

---

These classes compose the HBR module. They should be completely unaware of anything that starts with Norm

HBRConf

• Contains all configuration parameters for the HBR class
• Load and save functions (from/to JSON)
• Checks consistency of parameter configurations

HBR

• Concrete implementations of the fitting and prediction logic for hbr models
• Uses pymc
• receives and returns numpy arrays