the biggest one was that the Rs factor was used incorrectly: instead
of "surface volume of gas per volume of saturated reservoir oil" it is
defined as "surface volume of gas which a volume unit of surface oil
dissolves at reservoir pressure".
Besides this, there are a few smallish improvements like the oil
density being a function which can be called without having the
boilerplate objects (i.e., the fluid state and parameter cache
objects) instantiated.