On the ratio of energy produced to energy injected in SAGD: Long-term consequences of early stage operational decisions