Time-Myopic Go-Explore: Learning A State Representation for the Go-Explore Paradigm