For Wan 2.2 s2v passing a pose video does not seem to have any effect

### Custom Node Testing

- [x] I have tried disabling custom nodes and the issue persists (see [how to disable custom nodes](https://docs.comfy.org/troubleshooting/custom-node-issues#step-1%3A-test-with-all-custom-nodes-disabled) if you need help)

### Expected Behavior

I tried using a pose reference with Wan 2.2 s2v with the original code here and it just worked:
https://github.com/Wan-Video/Wan2.2
It worked even with a pose video without DWPose applied!

So I build a workflow to feed a DWPose video as pose video and assumed this would work.

### Actual Behavior

I got the s2v workflow running feeding a DWPose video as pose input but it seems to use the pose video input only for the first block (80 frames).

To render more than 80 frames I used the WanSoundImageToVideoExtend node and fed it the same pose video reference. This does NOT work. Instead the Pose input from the initial pose video frames are used again and again.

In the code of WanSoundImageToVideoExtend it seems it always only uses a couple of the first frames of the pose video and does not apply any frame offset in the node - which I think would be needed.

Is pose video support currently broken when using WanSoundImageToVideoExtend?

### Steps to Reproduce

Create a Wan 2.2 s2v Workflow and extend it using WanSoundImageToVideoExtend to render longer movie that just the initial 80 frames (that work).

### Debug Logs

```powershell
No Debug output hinting on any error. Seems to be a obvious bug in WanSoundImageToVideoExtend
```

### Other

_No response_