Re: Let an ordering index scan hand its ORDER BY value to the target list

Hello,

v13 attached, rebased on master (f8cabd1f175). It has two small
cleanups and no change in behavior:

* plannodes.h: the comment on IndexScan.indexorderbyexact now says
that setrefs.c replaces only float8 and float4 ORDER BY
expressions. (0002)

* createplan.c: create_append_child_plan() ends with
change_plan_targetlist() instead of repeating its logic. (0007)

The other six patches are the same as in v12. Each patch builds
without warnings and passes the regress, isolation and btree_gist
suites on its own. The series passes the full test suite with TAP
tests, the PG_TEST_EXTRA suites, and the running-server pass.

The commitfest started on the 1st, and the entry still has no
reviewer. Matthias, Oleg: if either of you has time, 0001 and 0002
are where to start. 0001 is a geometry fix that stands on its own:
the point-to-box distance is now exact, so GiST and SP-GiST kNN scans
return rows in the operator's order. 0002 adds the interface Oleg
tested. The rest builds on those two.

best.

-greg