Cleanup env/config, add LoRA training page, consolidate systemd service

feat: training API schemas, startup script model stacks, I2V prompt fix

Update AMD worker and Space deployment

Add AMD gallery training status UI

Prevent AI Toolkit installer from restarting services

Make AMD startup script idempotent

Add AI Toolkit training setup and live job progress

Add Nemoflix AMD backend and video stack

Add HF diffusers LoRA training stack

Fix startup script — actually runs ComfyUI test generation

Initial commit