[Feature] Support CogVideoX-5B-I2V

[LukeLIN-web]

https://huggingface.co/zai-org/CogVideoX-5b-I2V

Looks pretty good in https://research.nvidia.com/labs/dir/pbench/

[LukeLIN-web]

also https://huggingface.co/Wan-AI/Wan2.1-I2V-14B-720P

I think these are good at phsycial world benchmark, such as https://huggingface.co/datasets/nvidia/PBench, not 2D tasks.

Maybe we need to distinguish 3D task / 2D task?

[LukeLIN-web]

the best video model from https://lmarena.ai/leaderboard/text-to-video

https://huggingface.co/Wan-AI/Wan2.2-T2V-A14B

https://huggingface.co/genmo/mochi-1-preview

[LukeLIN-web]

The problem is that it may need submodule? Which one is easier to launch.