Re: DROP OWNED BY fails to clean out pg_init_privs grants

I wrote:
> A bigger problem though is that I think you are addressing the
> original complaint from the older thread, which while it's a fine
> thing to fix seems orthogonal to the failure we're seeing in the
> buildfarm. The buildfarm's problem is not that we're recording
> incorrect pg_init_privs entries, it's that when we do create such
> entries we're failing to show their dependency on the grantee role
> in pg_shdepend. We've missed spotting that so far because it's
> so seldom that pg_init_privs entries reference any but built-in
> roles (or at least roles that'd likely outlive the extension).

Here's a draft patch that attacks that. It seems to fix the
problem with test_pg_dump: no dangling pg_init_privs grants
are left behind.

A lot of the changes here are just involved with needing to pass the
object's owner OID to recordExtensionInitPriv so that it can be passed
to updateAclDependencies. One thing I'm a bit worried about is that
some of the new code assumes that all object types that are of
interest here will have catcaches on OID, so that it's possible to
fetch the owner OID for a generic object-with-privileges using the
catcache and objectaddress.c's tables of object properties. That
assumption seems to exist already, eg ExecGrant_common also assumes
it, but it's not obvious that it must be so.

regards, tom lane