[enhancement]: Support Loras created in Replicate

### Is there an existing issue for this?

- [X] I have searched the existing issues

### Contact Details

_No response_

### What should this feature add?

Add support for Loras in .safetensor format created in Replicate.com

Replicate.com is growing, I think mainly because it's very accessible and cheap to train custom models for virtually all kind of generative AI image models. And it just works! Besides you can also deploy your custom models on hardware and access them directly in Replicate or by using the API endpoint for your deployment. Compared to Google Colab notesbooks, most of those out there on the internet are broken or lack functions for training SDXL models.

However I'd like to run the trained Lora locally in Invoke. When I download my Lora from Replicate I get a .tar file that contains the lora.safetensors. When I try to load it in Invoke AI tho it says Unknown Lora type.

I tried the same thing in Automatic1111, which works.
It would be great to add this support to Invoke AI as well.

BR
William

### Alternatives

Automatic1111 can load Loras created with Replicate

### Additional Content

_No response_