Scalex: Scalability Exploration Of Multi-Agent Reinforcement Learning Agents In Grid-Interactive Efficient Buildings