intel/brw: Use urb_read_length instead of nr_attribute_slots to calculate VS first_non_payload_grf.

Makes sure the number of registers reserved for the payload matches
the size of the URB read, which prevents the VS shared function from
writing past the end of the register file on Xe3 with VRT enabled.

Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32664>

src/intel/compiler/brw_compile_vs.cpp

@@ -20,7 +20,7 @@ brw_assign_vs_urb_setup(fs_visitor &s)
    assert(s.stage == MESA_SHADER_VERTEX);
 
    /* Each attribute is 4 regs. */
-   s.first_non_payload_grf += 4 * vs_prog_data->nr_attribute_slots;
+   s.first_non_payload_grf += 8 * vs_prog_data->base.urb_read_length;
 
    assert(vs_prog_data->base.urb_read_length <= 15);