Strange behavior with missing column in SQL function

Here's the code that reproduces the behavior:
http://paste.debian.net/1035412/

I have already discussed this in the IRC channel but there doesn't seem to
be a consensus on whether this is a bug here's a brief transcript of
RhodiumToad's opinion:

> this isn't new, goes back to 9.1 at least
> basically, the error path in sql_fn_post_column_ref is a bit confused.
> seeing r.c it tries to resolve it as parameter.field, fails, and rather
than reporting the error directly as being a missing field, it just returns
with the reference unresolved
>then the outer parser code, having failed to resolve it as table.column
and having had the hook function not override it, reports it on the
assumption that it's a missing table
> so it's probably been this way for as long as named parameters have
worked in language sql
> msl09: as far as I can tell it's just giving the wrong error in an error
path, everything that's supposed to work does work
> msl09: but the error is definitely misleading

My question is "Is this a bug? Should it be reported?"