Update requirements.txt

Trying a new video backend to address errors:
video_reader_backend torchvision error, use torchvision as default, msg: 'video_fps'
https://github.com/QwenLM/Qwen3-VL/blob/a0d6248355509926085904f65c56a13a08f20bee/qwen-vl-utils/src/qwen_vl_utils/vision_process.py#L411

requirements.txt

@@ -5,6 +5,7 @@ pydantic==2.10.6
 gradio==5.23.2
 torch
 torchvision
+torchcodec
 git+https://github.com/huggingface/transformers
 accelerate
 qwen-vl-utils==0.0.14