On 2021-06-11T19:00:41+0300, Heikki Linnakangas wrote:
> We already have these same queries in the 'arrays' test against the
> 'array_op_test' table, though. It sure looks like a copy-paste error to me
> as well.
That's reason enough, but another reason is that I don't think GIN_CAT_NULL_KEY
is covered without this change.