Introduce a new ComfyUI plugin to integrate SGLang Diffusion. Adds core integration pieces (generator, model patcher, server API client), executor implementations for multiple model types (flux, qwen_image, zimage, native, anima), node mappings, utilities, example workflows, and tests. Includes README with installation/usage instructions and example workflow JSON files. This commit implements both server (HTTP) and integrated modes, model loading/patching logic, and basic LoRA/video/image API support.
