Lemonade v10.7 release and project organization update

In three linesLemonade v10.7 introduces omni-modal models (image gen/editing), a benchmarking tool (lemonade bench) comparing llama.cpp, FastFlowLM and vLLM, and extends multi-vendor support (CUDA, Vulkan). 19 contributors, 6 working groups with 4 led by non-AMDers. GPU acceleration on AMD, Apple Silicon, Nvidia, Intel.

Summary generated by Claude — human-verified