A Simple and Adaptive Learning Rate for FTRL in Online Learning with Minimax Regret of \Theta(T {2/3}) and its Application to Best-of-Both-Worlds