This function can be used to initialize a big integer to either the upper
or lower limit of a 2s-complement integer. Note that the result is still
in sign-magnitude representation, though in order to convert it into twos
complement all one has to do is take the absolute value.