Natural simple float on integer class. However, we should use another clang -msoft-float option first when practical use. Fixed bug around bit shifts. This library is moderately withered without zeroFix, matrix pow/exp/log functions.
