[Bugfix] Fix Qwen3-Omni audio-input mean_e2el_ms regression (#8306)

Signed-off-by: psv666 <2693925048@qq.com>
Co-authored-by: amy-why-3459 <wuhaiyan17@huawei.com>

vllm_omni/model_executor/models/qwen3_omni/qwen3_omni_moe_thinker.py

@@ -678,6 +678,8 @@ class Qwen3OmniMoeThinkerProcessingInfo(Qwen2AudioProcessingInfo, Qwen2_5_VLProc
             spatial_merge_size=self.get_hf_config().vision_config.spatial_merge_size,
             target_sr=feature_extractor.sampling_rate,
             target_channels=1,
+            # Retain PyAV for long-audio latency after vLLM 0.30 changed the default.
+            audio_resample_method="pyav",
             expected_hidden_size=self._get_expected_hidden_size(),
         )