[Misc] Qwen-Omni support offline inference with local files

[SamitHuang]

PLEASE FILL IN THE PR DESCRIPTION HERE ENSURING ALL CHECKLIST ITEMS (AT THE BOTTOM) HAVE BEEN CONSIDERED.

Purpose

Resolve #161

Test Plan

cd examples/offline_inference/qwen3_omni
# cd examples/offline_inference/qwen2.5_omni

Then run

# Use local video file
python end2end.py --query-type use_video --video-path /path/to/video.mp4

# Use local image file
python end2end.py --query-type use_image --image-path /path/to/image.jpg

# Use local audio file
python end2end.py --query-type use_audio --audio-path /path/to/audio.wav

Test Result

--------------------------------
[Stage-0] Received batch size=1, request_ids=[0]
--------------------------------
[Stage-0] Generate done: batch=1, req_ids=[0], gen_ms=7178.5
[Stage-1] Max batch size: 1
--------------------------------
[Stage-1] Received batch size=1, request_ids=[0]
--------------------------------
(EngineCore_DP0 pid=1750793) /home/public/yx/vllm-omni/vllm_omni/worker/gpu_model_runner.py:207: UserWarning: The given NumPy array is not writable, and PyTorch does not support non-writable tensors. This means writing to this tensor will result in undefined behavior. You may want to copy the array to protect its data or make it writable before converting it to a tensor. This type of warning will be suppressed for the rest of this program. (Triggered internally at /pytorch/torch/csrc/utils/tensor_numpy.cpp:203.)
(EngineCore_DP0 pid=1750793)   info_dict[k] = torch.from_numpy(arr)
[Stage-1] Generate done: batch=1, req_ids=[0], gen_ms=12316.4
[Stage-2] Max batch size: 1
--------------------------------
[Stage-2] Received batch size=1, request_ids=[0]
--------------------------------
(EngineCore_DP0 pid=1751740) INFO:vllm_omni.model_executor.models.qwen2_5_omni.qwen2_5_omni:Currently, we do not use the chunked process, we only use the token2wav.process_chunk for the whole sequence. The stream mode will be implemented in the future.
[Stage-2] Generate done: batch=1, req_ids=[0], gen_ms=4843.1
INFO:vllm_omni.entrypoints.omni_llm:[Summary] {'e2e_requests': 1, 'e2e_total_time_ms': 24620.004653930664, 'e2e_sum_time_ms': 24619.704723358154, 'e2e_total_tokens': 0, 'e2e_avg_time_per_request_ms': 24619.704723358154, 'e2e_avg_tokens_per_s': 0.0, 'wall_time_ms': 24620.004653930664, 'final_stage_id': 2, 'stages': [{'stage_id': 0, 'requests': 1, 'tokens': 89, 'total_time_ms': 7263.629674911499, 'avg_time_per_request_ms': 7263.629674911499, 'avg_tokens_per_s': 12.252827303050026}, {'stage_id': 1, 'requests': 1, 'tokens': 1082, 'total_time_ms': 12408.075332641602, 'avg_time_per_request_ms': 12408.075332641602, 'avg_tokens_per_s': 87.20127586214848}, {'stage_id': 2, 'requests': 1, 'tokens': 0, 'total_time_ms': 4851.06635093689, 'avg_time_per_request_ms': 4851.06635093689, 'avg_tokens_per_s': 0.0}], 'transfers': [{'from_stage': 0, 'to_stage': 1, 'samples': 1, 'total_bytes': 36376783, 'total_time_ms': 47.77932167053223, 'tx_mbps': 6090.799404954347, 'rx_samples': 1, 'rx_total_bytes': 36376783, 'rx_total_time_ms': 70.0995922088623, 'rx_mbps': 4151.44018431549, 'total_samples': 1, 'total_transfer_time_ms': 118.60966682434082, 'total_mbps': 2453.545919077472}, {'from_stage': 1, 'to_stage': 2, 'samples': 1, 'total_bytes': 3265, 'total_time_ms': 0.21910667419433594, 'tx_mbps': 119.21133893362351, 'rx_samples': 1, 'rx_total_bytes': 3265, 'rx_total_time_ms': 0.03528594970703125, 'rx_mbps': 740.2379762162162, 'total_samples': 1, 'total_transfer_time_ms': 0.8287429809570312, 'total_mbps': 31.517612336018413}]}
[rank0]:[W1202 10:13:07.573033067 ProcessGroupNCCL.cpp:1538] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
[rank0]:[W1202 10:13:07.603865240 ProcessGroupNCCL.cpp:1538] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
[rank0]:[W1202 10:13:07.616500689 ProcessGroupNCCL.cpp:1538] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
Request ID: 0, Text saved to output_audio/00000.txt
Request ID: 0, Saved audio to output_audio/output_0.wav

---

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan, such as providing test command.
• The test results, such as pasting the results comparison before and after, or e2e results
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model.
• (Optional) Release notes update. If your change is user facing, please update the release notes draft.

BEFORE SUBMITTING, PLEASE READ https://github.com/vllm-project/vllm-omni/blob/main/CONTRIBUTING.md (anything written below this line will be removed by GitHub Actions)

[DarkLight1337]

cc @Isotr0py @ywang96 as with the other PR, let's try to upstream support for loading from local path