Physics-based, data-driven, and hybrid methods to predict the energy requirements for cargo heating: a case study for chemical tankers