Component-wise gradient boosting for analysis of multiply
imputed datasets. Implements the algorithm Boosting after Multiple
Imputation (MIBoost), which enforces uniform variable selection across
imputations and provides utilities for pooling. Includes a cross-validation
workflow that first splits the data into training and validation sets and
then performs imputation on the training data, applying the learned
imputation models to the validation data to avoid information leakage.
Supports Gaussian and logistic loss. Methods relate to gradient boosting
and multiple imputation as in Buehlmann and Hothorn (2007)