How to merge duplicate columns in pyspark