[Misc] Move some multimodal utils to modality-specific modules (#11494)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>

vllm/multimodal/audio.py

@@ -1,3 +1,5 @@
+from typing import Any
+
 import numpy as np
 import numpy.typing as npt
 
@@ -26,6 +28,16 @@ class AudioPlugin(MultiModalPlugin):
             "There is no default maximum multimodal tokens")
 
 
+def try_import_audio_packages() -> tuple[Any, Any]:
+    try:
+        import librosa
+        import soundfile
+    except ImportError as exc:
+        raise ImportError(
+            "Please install vllm[audio] for audio support.") from exc
+    return librosa, soundfile
+
+
 def resample_audio(
     audio: npt.NDArray[np.floating],
     *,