VRAM Out-of-Memory Error When Using LoRA with Qwen_image_edit in ComfyUI v0.7

### 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

After updating ComfyUI to version 0.7, a GPU VRAM out-of-memory error occurs when using Qwen_image_edit with LoRA enabled. The error message is displayed as follows:
KSamplerAllocation on deviceThis error means you ran out of memory on your GPU.

### Actual Behavior

This issue also occurs when using the standard official workflow "mage_qwen_image_edit_2511".

### Steps to Reproduce

[Error reporting.txt](https://github.com/user-attachments/files/24402526/Error.reporting.txt)

### Debug Logs

```powershell
The issue is completely resolved once LoRA is disabled.
```

### Other

_No response_