[PATCH] login: Add back libutil as an empty library

[Florian Weimer]

* Stafford Horne via Libc-alpha:

> There are several packages like sysvinit and buildroot that expect
> -lutil to work.  Rather than impacting them with having to change
> the linker flags provide an empty libutil.a.

This produces a libutil.so.1 as well, right?

I think we should still prevent that.  It's not clear to me what the
proper approach would be, though.  We have a bunch of existing
examples: libmcheck.a, libg.a.  But those do not use the usual library
mechanism AFAICS.

Maybe this will work?

libutil-inhibit-o = $(filter-out .o,$(object-suffixes))

And do not set libutil-routines and libutil-shared-only-routines for
new architectures.