radeonsi: upload constants into VRAM instead of GTT

radeonsi: upload constants into VRAM instead of GTT

This lowers lgkm wait cycles by 30% on VI and normal conditions.
The might be a measurable improvement when CE is disabled (radeon)
or under L2 thrashing.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>