Pyspark Dataframe Drop Duplicate Columns