pub fn train_test_split_indices( n_samples: usize, test_size: f64, shuffle: bool, seed: Option<u64>, ) -> UtilsResult<(Vec<usize>, Vec<usize>)>
Split indices into train/test sets