fix: disable fast_inference (vLLM not installed) in training/evaluate.py

Update CERNenv Space