Do we need deduplicate multikey sort logics in GPDB and Postgres upstream?

Currently, GPDB supports multi key sort already, but for PG 13, incremental sort is also introduced. Should we deduplicate the multi key sort logics which are similar with each other?

The issue names no files, tests, or entry points. Start by comparing GPDB's existing multi-key sort logic with PostgreSQL 13's incremental-sort implementation, then determine whether deduplication is technically appropriate and define the scope and completion criteria for the proposed refactor.

Written by the indexing model from the issue text.