In-Context LoRA

[filipstrand]

For this feature I did not follow any reference implementation, so I hope it is correct (but does look promising on some initial tests). Would be cool to include in v.0.6.0 which should be released soon (otherwise leave it for next release along with some other editing techniques).

Example:

In this set of two images, a bold modern typeface with the brand name 'DEMA' is introduced and is shown on a company merchandise product photo; [IMAGE1] a simplistic black logo featuring a modern typeface with the brand name 'DEMA' on a bright light green/yellowish background; [IMAGE2] the design is printed on a green/yellowish hoodie as a company merchandise product photo with a plain white background.

Uses the LoRA adapters from https://huggingface.co/ali-vilab/In-Context-LoRA

Left to do:

• (DONE) Some cleanup
• (DONE - Just accept a regular prompt for widest possible options) Think about the best kind of interface for this feature (e.g maybe have 3 separate input prompts - global, IMAGE1 and IMAGE2...the general setup seem to map a set of input images to a set of output ones, so it could be too restrictive to limit this to 1-in 1-out...)
• (DONE) Automatically download the relevant LoRAs files from HF Hub, like we do for controlnet files etc.
• (DONE) Write docs, include examples etc.

Also updates some of the existing interface for inputting images
init_image -> image
init_image_strength -> image_strength

Note to self: remember to document this in release notes