feat: add per-model FP8 layerwise casting for VRAM reduction#8945

Pfannkuchensack wants to merge 5 commits intoinvoke-ai:maininvoke-ai/InvokeAI:mainfrom

Pfannkuchensack:feature/fp8-layerwise-castingPfannkuchensack/InvokeAI:feature/fp8-layerwise-castingCopy head branch name to clipboard