Add MOSS-Image edit LoRA (rank 64, bf16)
Rehosts the instruction-edit LoRA from https://huggingface.co/chinoll/Moss-Image-Edit (revision 3be9c40, adapter_model.safetensors, sha256 17a48b8c605dcffdedb92a2b887d18cf316505be90802717669fb6556098dcdf, 297,382,216 bytes, bf16, rank 64 / alpha 64).
File is uploaded byte-identical to upstream; only the name changes to follow the repo convention (loras/<model>_<purpose>_lora_rank_<N>_<dtype>.safetensors).
Verified in ComfyUI (0.42.0 + PR #17296): loads with LoraLoaderModelOnly and pairs with the Text Encode MOSS-Image Edit node using edit_format = Moss-Image-Edit; real runs on a 4090-class box at 30 steps / cfg 4.0 / euler + normal kept identity, pose and framing.
README gains the upstream source link, the file in the models tree, and a short LoRAs section covering this edit LoRA and the existing color/depth control adapter.
The loader notes here are handy, since the LoRAs section didn't make it into the README.