Add img2img and inpaint stable diffusion ORT pipelines (#1121)
* add img2img pipeline
* fix style
* fix import
* fix style
* add diffusers extra
* fix num_inference_steps
* add test
* add test for diffusers pipeline loading compatibility
* Modify config to keep the resulting model compatible with diffusers pipelines
* fix style
* add inpaint pipeline
* add test
* fix style
* remove redudant test
* add tests
* fix style
* add tests
* save config
* add subcomponent config saving
* set VAE scaling factor in pipeline
* move ORTModelTestMixin
* add latent output type
* fix dosctrings format
* fix style
* update ORTModelTestMixin
* fix callbacks and add test
