All Roads Lead to Rome? Exploring Representational Similarities Between Latent Spaces of Generative Image Models