So I checked in Christian Limpach's bug fixes for SCSI and Turbo
support.  I also made some changes to allow the same boot block and
kernel to work on both turbo and non-turbo machines, so there is no
longer a need to build two sets.  I also fixed some problems in the
console driver.

I've done some testing, and can verify that with these changes I was
able to netboot both a mono slow slab and a color turbo slab.
(Unfortunately, my mono turbo slab is not here, and my color slow slab
seems to be defective -- it can't pass self-test.)

There are some caveats, though:

1) I didn't have any SCSI drives handy to test, so the SCSI support
   may or may not work.  Please let me know either way, but especially
   if it does.

2) On my color turbo, with the SCSI driver enabled, the kernel wedges
   in a loop printing `esp0: SCSI bus reset' during autoconfig.  I
   haven't looked at this.

3) On the turbo, I get `xe0: discarding oversize frame (len=1519)'
   periodically (always with the same `len'), and network access seems
   to be slower than I would expect.  It definitely feels slow than
   the non-turbo slab.  I suspect something is still wrong here.

4) On the turbo, the boot block sometimes (well, really more often
   than not) aborts loading with `enintr: timed out'.

That said, it's a step in the right direction, but it's not completely
working yet.