[Data] Add include_row_hash to read_parquet

[wingkitlee0]

Description

Having a row hash column will be useful for checkpointing, e.g., in Ray Data pipeline or Ray Train use cases

Use case

Ray Data (read-map-write): #59409
requires an existing "id" column. This PR will automatically generate a row_hash column that simplifies that.

Ray Train:

• many users are implementing their own stateful dataloader (as indicated in Ray Summit..)
• Anyscale has their own mid-epoch resumption: https://docs.anyscale.com/runtime/mid-epoch-resumption
  • and generate_id_column option.