Clean notebook outputs and add Colab warning note

Finalizing CodeArena RL Benchmark: frontend improvements, GRPO training scripts, and cleaned environment

fix: removed invalid openenv-py package from notebook install cell

feat: use m-a-p/Code-Feedback dataset for GRPO training

Complete all tasks: Adaptive curriculum, GRPO, React frontend, LLM-as-a-judge