scikit-covtest datasets: MNIST and TCGA

Benchmark datasets for the scikit-covtest Python library, a toolkit for classical and high-dimensional covariance matrix hypothesis testing. This package contains preprocessed .npz files for MNIST (784-dimensional digit images across train and test splits) and TCGA (pan-cancer gene expression profiles with associated sample metadata and labels). These datasets are automatically retrieved and cached via pooch on first use in scikit-covtest for evaluation, benchmarking, and demonstration of high-dimensional statistical tests.