Add banded ridge regression

[Hugo-W]

Description

Implement banded ridge regression (non-spherical priors) for TRF estimation, as described in the Gallant Lab's tikreg library.

Tasks

• Implement banded ridge regression with non-spherical priors.
• Extend the data to transform any Tikhonov regularization scheme into an OLS equation.
• Integrate with existing TRFEstimator class.

Mathematical Background

Banded ridge regression allows for different regularization strengths for different features or groups of features, enabling more flexible modeling of the TRF.

Acceptance Criteria

• Banded ridge regression is available as an option in TRFEstimator.
• Backward compatibility with existing TRFEstimator usage is maintained.
• Performance is comparable to existing regularization methods.

Priority

Medium