Re: pg_dump: assert failure sorting casts/transforms

> here is v3 version of the patch addressing all nit-picks

Thanks for the v3 patch. It addresses my earlier comments, and the fix
and regression coverage look good to me. I have no other comments.

One optional follow-up thought: the current tests exercise the
assertion-failure case well. Order-aware checks could additionally
verify the OID-independent ordering property and help detect future
regressions to an OID fallback. I realize the earlier natural-key
ordering changes do not consistently include explicit output-order
checks either, so I do not think this should delay the current patch
or require a v4. If there is interest, I can investigate a separate
follow-up patch for suitable coverage of this case and the earlier
OID-independence cases.

Since the issue also affects REL_19_STABLE, I think the fix should be
applied there along with master. No older back-branches appear to be
affected.

Best Regards,
Nitin Jadhav
Azure Database for PostgreSQL
Microsoft