Review of patch renaming constraints

> Make check passed.  Patch has tests for rename constraint.
>
> Most normal uses of alter table ... rename constraint ... worked normally.  However, the patch does not deal correctly with constraints which are not inherited, such as primary key constraints:

That appears to be because creating a primary key constraint does not
set pg_constraint.conisonly correctly.  This was introduced recently
with noninherited check constraints.

> And primary keys are anyways not inherited. So why is the conisonly
> field interfering in rename? Seems quite orthogonal to me.

In the past, each kind of constraint was either always inherited or
always not, implicitly.  Now, for check constraints we can choose what
we want, and in the future, perhaps we will want to choose for primary
keys as well.  So having conisonly is really a good step into that
future, and we should use it uniformly.