Tra-MoE: Learning Trajectory Prediction Model from Multiple Domains for Adaptive Policy Conditioning