An advantage in using an ID Col would be that with RRN() you could
lose some records if your file has been created with REUSEDLT(*YES),
as a new record could be inserted in an old RRN. ID Columns (AFAIK)
would not have that issue.
I wonder though, what would all of the overhead of an IDcol achieve
that a RRN() could not, for the same scenario? The relative row
number is a similarly arbitrary value, as compared to an identity.