Hive-partition columns are not correctly stored in index content when refresh-incremental is called

[apoorvedave1]

Describe the issue

When incremental refresh is called on hive-partitioned data, if partitoin columns are part of index columns (indexed/included), they are not picked by the refresh call and are filled with nulls.

To Reproduce

1. create hive partitioned data. e.g. df.write.partitionBy("c1").parquet...
2. create index where c1 is used in either indexed or included columns
3. append new data to source
4. hs.refresh("index", "incremental")
5. check index data. 'c1' column will be filled with 'nulls' for all appended data.

Expected behavior

'c1' column should contain proper partition values instead of nulls