add qwq testcase

[ck-hw-1018]

What this PR does / why we need it?

This PR adds a qwq case for nightly test for qwen-qwq on A3 ,we need test them daily

Does this PR introduce any user-facing change?

no

How was this patch tested?

by running the test

• vLLM version: v0.11.0rc3
• vLLM main: vllm-project/vllm@c9461e0

[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 PR adds a new end-to-end test for the Qwen/QwQ-32B model. The overall structure is good, but I've found an area for improvement in how the server arguments are constructed. The current implementation is brittle and hard to maintain. I've provided a suggestion to make it more robust and readable.

[gemini-code-assist]

The current method of constructing server_args by defining a default list and then modifying it with list.remove() is brittle and can lead to runtime errors. If the initial list is changed, the remove() calls might fail with a ValueError. Additionally, the long JSON string for compilation_config is duplicated, making the code harder to maintain.

It's better to build the argument list conditionally from common and mode-specific parts. This approach is more robust, readable, and avoids duplicating configuration strings.

    server_args = [
        "--tensor-parallel-size",
        str(tp_size), "--port",
        str(port), "--max-model-len", "36864", "--max-num-batched-tokens",
        "36864", "--block-size", "128", "--trust-remote-code",
        "--gpu-memory-utilization", "0.9",
    ]
    if mode == "single":
        server_args.extend([
            "--additional-config",
            '{"ascend_scheduler_config":{"enabled":true}}',
            "--enforce-eager",
        ])
    else:  # aclgraph
        server_args.extend([
            "--compilation_config",
            '{"cudagraph_mode":"FULL_DECODE_ONLY", "cudagraph_capture_sizes": [1, 8, 24, 48, 60]}',
        ])
    server_args.extend([
        "--reasoning-parser", "deepseek_r1", "--distributed_executor_backend",
        "mp"
    ])

[ck-hw-1018]

https://github.com/vllm-project/vllm-ascend/actions/runs/18800163569/job/53646637879?pr=3757

[jiangyunfan1]

LGTM