I just finished updating the code with an assembly version of a hot function (finding the leading "1" bit position) and updated the blog to reflect the improvem

I just finished updating the code with an assembly version of a hot function (finding the leading "1" bit position) and updated the blog to reflect the improvements. In general I saw a 3-40% performance increase using the assembly version, which I borrowed from math/big/arith_*.s. The reason for the wide range is because it depended on the size of the delta integers being processed.