I represent positive integers of more than 30 bits (ie, >= 1073741824).
These values are beyond the range of SmallInteger, and are encoded here as an array of 8-bit digits. Care must be taken, when new values are computed, that any result that COULD BE a SmallInteger IS a SmallInteger (see normalize).
See superclasses Integer and Number for common methods
Note that the bit manipulation primitives, bitAnd:, bitShift:, etc., = and ~= run without failure (and therefore fast) if the value fits in 32 bits. This is a great help to the simulator.