[Feature] Synchronize vLLM mrope modifications, support Qwen2.5-OMni-7B Thinker.

[Gong-xxx]

…7B Thinker module

What this PR does / why we need it?

Synchronize vLLM mrope modifications, support Qwen2.5-OMni-7B thinker module.

Does this PR introduce any user-facing change?

support audio mrope

How was this patch tested?

Related: #967

[Yikun]

@Gong-xxx Thanks for your contributions!

• Cleanup CI error according: https://vllm-ascend.readthedocs.io/en/latest/developer_guide/contributing.html#prepare-environment-and-build , such as run format.sh
• Please also update commit message: How was this patch tested?
• Please also add the test: https://github.com/vllm-project/vllm-ascend/blob/main/tests/singlecard/test_offline_inference.py
  • You can add https://modelscope.cn/models/Qwen/Qwen2.5-Omni-3B as test
• Any plan to support in V1?

[wangxiyuan]

++ for V1 engine. V0 will not be supported in the future.

[Yikun]

We don't need this anymore.

pip install qwen-omni-utils -U
vllm serve Qwen/Qwen2.5-Omni-7B --enforce-eager

curl http://localhost:8000/v1/chat/completions \
    -H "Content-Type: application/json" \
    -d '{
    "messages": [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": [
        {"type": "image_url", "image_url": {"url": "https://modelscope.oss-cn-beijing.aliyuncs.com/resource/qwen.png"}},
        {"type": "audio_url", "audio_url": {"url": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen2.5-Omni/cough.wav"}},
        {"type": "text", "text": "What is the text in the illustrate ans what it the sound in the audio?"}
    ]}
    ]
    }'

{"id":"chatcmpl-9ddb4f86a49343f989fcffeb8ac28241","object":"chat.completion","created":1752334669,"model":"Qwen/Qwen2.5-Omni-7B","choices":[{"index":0,"message":{"role":"assistant","reasoning_content":null,"content":"The text in the image is \"TONGYI Qwen.\" The sound in the audio is a cough.","tool_calls":[]},"logprobs":null,"finish_reason":"stop","stop_reason":null}],"usage":{"prompt_tokens":156,"total_tokens":180,"completion_tokens":24,"prompt_tokens_details":null},"prompt_logprobs":null,"kv_transfer_params":null

I test it in v0.9.2rc1, so close this.