Allow Setting `original_config_file` for Diffusers Backend

Is your feature request related to a problem? Please describe.

I encountered an error while using diffusers backend for text-to-image service. Loading model failed due to error:
ERR Server error error="could not load model (no success): Unexpected err=OSError('DreamShaperXL_Turbo_v2_1.safetensors does not appear to have a file named model_index.json.'), type(err)=<class 'OSError'>".
According to the discussion https://github.com/huggingface/diffusers/issues/7209#issuecomment-1978067476 from_single_file function need the argument original_config_file in a offline environment.

Describe the solution you'd like

Please improve backend/python/diffusers/backend.py, allow pass in original_config_file while loading model from local.

Describe alternatives you've considered

n/a

Additional context

https://github.com/huggingface/diffusers/issues/7209

Start in backend/python/diffusers/backend.py and trace the local model-loading path that calls Diffusers' from_single_file function. Check how a local original_config_file could be supplied, then verify that loading the reported DreamShaperXL_Turbo_v2_1.safetensors case works offline with that argument.