Number representation

The number representation we chose was the 18-bit (or 16-bit, depending on the part of the meta-pipeline) fixed point. Originally we attempted to implemented a simpler version of IEEE floating point specification (IEEE754) - specifically without infinity and NaN support - so that number conversions from software on the NIOS II could be as easy as a cast in C. However, the idea was scrapped in the interests of time, since a prior lab in the class already introduced us to the 18-bit fixed point format.