I am saving a parquet file with spark where one of the columns is decimal. Physical type of this column becomes INT32 and INT64 based on its precision. Then, when I read the parquet file with ...