結果 : spark dataframe remove duplicate rows based on one column