[BugFix] Fix input parameter bug of dispatch_gmm_combine_decode[RFC: issue 5476]

[wangqiankun13]

What this PR does / why we need it?

In PR 5040, the dispatch_gmm_combine_decode operator was configured with an incorrect global_bs parameter. This PR is to fix the bug.

The global_bs provided as input should have the same meaning as in the moe_distributed_dispatch operator, specifically: (the maximum batch size across all cards) * (expert parallel world size).
However, the implementation incorrectly used the variable max_num_tokens, which does not account for tensor parallelism. This error likely resulted in an unnecessarily large (overestimated) value.

More info about this operator, please refer to RFC: issue #5476

Does this PR introduce any user-facing change?

No

How was this patch tested?

Acc
test qwen3-235b eplb on a single A3 node(ep16),
with dispatch_gmm_combine_decode

nic_name="xxxx"
local_ip="xxx.xxx.xxx.xxx"

export HCCL_IF_IP=$local_ip
export GLOO_SOCKET_IFNAME=$nic_name
export TP_SOCKET_IFNAME=$nic_name
export HCCL_SOCKET_IFNAME=$nic_name

export VLLM_ASCEND_ENABLE_FUSED_MC2=2
echo "VLLM_ASCEND_ENABLE_FUSED_MC2=${VLLM_ASCEND_ENABLE_FUSED_MC2}"

export HCCL_OP_EXPANSION_MODE="AIV"
export HCCL_BUFFSIZE=512
export PYTORCH_NPU_ALLOC_CONF=expandable_segments:True
export OMP_PROC_BIND=false
export OMP_NUM_THREADS=10
export LD_PRELOAD=/usr/lib/aarch64-linux-gnu/libjemalloc.so.2:$LD_PRELOAD
vllm serve /dataset/Qwen3-235B-A22B-Instruct-2507-w8a8-QuaRot/ \
        --served-model-name "qwen" \
        --host 0.0.0.0 \
        --port 8004 \
        --async-scheduling \
        --tensor-parallel-size 4 \
        --data-parallel-size 4 \
        --max-num-seqs 64 \
        --max-model-len 40960 \
        --max-num-batched-tokens 16384 \
        --gpu-memory-utilization 0.9 \
        --enable-expert-parallel \
        --no-enable-prefix-caching \
        --quantization "ascend" \
        --trust-remote-code \
        --speculative_config \
        '{
            "method": "eagle3",
            "model": "/dataset/Qwen3-235B-A22B-Instruct-2507-speculator-eagle3/",
            "num_speculative_tokens": 2
        }' \
        --compilation-config '{"cudagraph_mode": "FULL_DECODE_ONLY"}' \
        2>&1 | tee qwen3_235b_eagle3.log

dataset	version	metric	mode	vllm-api-stream-chat
aime2024	604a78	accuracy	gen	80.00

• vLLM version: v0.13.0
• vLLM main: vllm-project/vllm@11b6af5

[github-actions]

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

• A PR should do only one thing, smaller PRs enable faster reviews.
• Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
• Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.

[gemini-code-assist]

Code Review

This pull request addresses a bug in the dispatch_gmm_combine_decode function by correcting the global_bs input parameter. The change replaces the usage of fused_global_bs with the correctly calculated global_bs, ensuring consistency with other related operator calls. The corresponding removal of the fused_global_bs attribute from TokenDispatcherWithMC2 is a clean and correct modification. The fix appears to be well-targeted and sound.