Replace torchvision.io.read_video with a PyAV-based shim

torchvision>=0.26 (auto-installed alongside torch 2.11.0) removed the
public read_video/write_video API entirely from torchvision.io in OSS
builds, breaking TARO onset extraction with ImportError. av (PyAV) was
already a pinned dependency, so decode frames directly with it instead.

TARO/onset_util.py

@@ -1,12 +1,27 @@
+import av
 import torch
 import torch.nn as nn
-from torchvision.io import read_video
 import os
 from einops import rearrange
 import torchvision.transforms as transforms
 from cavp_util import reencode_video_with_diff_fps
 
 
+def read_video(path, pts_unit="sec", output_format="TCHW"):
+    """Minimal PyAV-based stand-in for torchvision.io.read_video (removed from
+    the public torchvision.io API in torchvision>=0.26). Returns a (T, C, H, W)
+    uint8 frame tensor to match the subset of the old API this module relies on."""
+    container = av.open(path)
+    frames = [
+        torch.from_numpy(frame.to_ndarray(format="rgb24"))
+        for frame in container.decode(video=0)
+    ]
+    container.close()
+    video = torch.stack(frames, dim=0) if frames else torch.empty(0, 0, 0, 3, dtype=torch.uint8)
+    video = video.permute(0, 3, 1, 2).contiguous()  # (T, H, W, C) -> (T, C, H, W)
+    return video, None, None
+
+
 def extract_onset(video_path, onset_model, tmp_path, device="cuda"):
     """Extract onset features from video using a pre-trained onset detection model."""
     # Preprocess the video frames