A natural lead‐in approach to response‐adaptive allocation for continuous outcomes