Multi-agent Path Finding for Timed Tasks using Evolutionary Games