nir/lower_int64: Use the core NIR lowering framework