When reading in multiple parquet files into a dataframe, it seems to evaluate per parquet file afterwar