What’s the best way to read in parquet files (I have thousands) and start processing them into datasets? I’m currently trying to follow the example from: https://github.com/zero-one-group/geni-performance-benchmark/blob/master/dataset/src/dataset/optimised_by_chris.clj