fix(lora): patch non-diffusers Wan LoRA converter for ComfyUI and Civitai key compatibility

feat(turbo): add native 4-step and 8-step LightX2V distill Turbo LoRA acceleration

fix(deps): add peft dependency required for diffusers load_lora_weights

fix(gradio): align gradio with sdk 6.20.0 and restore standard action handlers

perf(zerogpu): pre-load model pipeline on CPU at startup and increase GPU timeout quota

fix(model_repo): use official verified public Wan2.2 Hugging Face repo IDs

fix(diffusers): import WanImageToVideoPipeline, WanPipeline, and DiffusionPipeline safely

fix(gradio): remove unsupported ssr kwarg and set GRADIO_SSR=False in env

fix(zerogpu): use dynamic *args in get_duration and disable experimental ssr

feat: upgrade default model to Wan 2.2 MoE suite and configure Civitai API key

fix(gradio): apply safe monkey-patch for gradio_client bool schema bug and set api_name=False

fix(gradio): fix OpenAPI bool schema crash and add Base NSFW Adapter slot with Civitai 1307155

docs: add push instructions and secret config guide in comandos.md

feat: initial commit for Wan 2.1 / 2.2 Studio Space with ZeroGPU and Multi-LoRA