[ENH] Add cross-validation function for brain vs covariate comparison

[chiuhoward]

Add cross-validation function for brain vs covariate comparison

Summary

This PR adds a new function compare_brain_covariates_cv that performs cross-validation to compare brain features (with optional covariate regression) against covariates-only models.

Changes

• Added IdentityTransformer class for no-op transformations in pipelines
• Added compare_brain_covariates_cv function that:
  • Compares brain features vs covariates across CV folds
  • Optionally uses CovariateRegressor to remove covariate effects from brain data
  • Returns detailed metrics for each fold including R², MAE, and model coefficients
• Added comprehensive tests for both new functions
• Updated __all__ exports

Usage

from afqinsight.pipeline import compare_brain_covariates_cv

results = compare_brain_covariates_cv(
    X_brain, X_covariates, y,
    regress_covariates=True,  # Use CovariateRegressor
    n_repeats=5, n_splits=10
)