Patch to master for native Nucleus support

THIS IS OUTDATED. DO NOT USE. USE THE PR HERE:

https://github.com/Comfy-Org/ComfyUI/pull/13471

### Feature Idea

Support Nucleus.

[nucleus_support_checkout.patch](https://github.com/user-attachments/files/26862386/nucleus_support_checkout.patch)

[nucleus_fp8_test.json](https://github.com/user-attachments/files/26862390/nucleus_fp8_test.json)

### Existing Solutions

None.

### Other

I'm a professional developer, but this code was written with AI (GLM-5.1 and ChatGPT 4.2 codex) so I don't know exactly what state it's in, which is why I'm not submitting a full PR. I can say that it works.

FP8 quants of the model and text encoder will appear here when they're done uploading:

[e-n-v-y/Nucleus-Image-FP8-e4m3fn-scaled](https://huggingface.co/e-n-v-y/Nucleus-Image-FP8-e4m3fn-scaled)