r600: Fix lower-to-scalar on TTN path

r600: Fix lower-to-scalar on TTN path

We must limit what alu ops are scalazized, and since we now
run finalize_nir we also have to make sure not to try to
scalarize backend specific instructions that are introduced
there.

While we touch the code, also remove the vectorization, it is
no longer helping.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7747

Fixes: a81c50a2 (r600/sfn: implement finalize_nir)

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19892>