History vs. chance in the evolution of steroid signaling