[Issue]: The `bias` is not used consistently in `gemm_a8w8_bpreshuffle`

Problem Description

The gemm_a8w8_bpreshuffle changed in this commit caf1e1914c28efac71697d2046bd9ebef7b7ec6a. it changed the API of gemm_a8w8_bpreshuffle and introduced bias term.
The bias is not used consistently.
bias is only used with asm kernel gemm_a8w8_ASM but not ck kernels, this makes the API interface usage confusing. Since gemm_a8w8_bpreshuffle is a tuning API interface, it is hard to know if when bias is passed into this API, will the bias be applied.

def gemm_a8w8_bpreshuffle(
    XQ: Tensor,
    WQ: Tensor,
    x_scale: Tensor,
    w_scale: Tensor,
    bias: Optional[Tensor] = None,
    dtype=torch.float16,
    check=False,
):
    assert dtype in [
        torch.bfloat16,
        torch.float16,
    ], f"Output {dtype=} is currently not supported in gemm_a8w8"
    m = XQ.shape[0]
    n = WQ.shape[0]
    k = XQ.shape[-1]

    ck_config = get_CKGEMM_config(m, n, k, "a8w8_bpreshuffle_tuned_gemm.csv")
    if (
        ck_config is None
        and dtype == dtypes.bf16
        and bias is not None
        and WQ.dtype != dtypes.i8
    ):
        res = gemm_a8w8_ASM(XQ, WQ, x_scale, w_scale, bias, dtype=dtype, check=check)
        if res is not None:
            return res
    Y = torch.empty(m, n, dtype=dtype, device=XQ.device)
    return gemm_a8w8_bpreshuffle_ck(XQ, WQ, x_scale, w_scale, Y)

https://github.com/ROCm/aiter/blob/2b083e30d524f0bbd4921e2922103322ac3efd04/aiter/ops/gemm_op_a8w8.py#L214

Operating System

Ubuntu 22.04.4 LTS (Jammy Jellyfish)

CPU

AMD EPYC 9654 96-Core Processor

GPU

AMD Instinct MI300X

ROCm Version

ROCm 6.4.1

ROCm Component

No response

Steps to Reproduce

No response

(Optional for Linux users) Output of /opt/rocm/bin/rocminfo --support

No response

Additional Information

No response