Implements new look-up table based method for reflect_8, reflect_16, and reflect_32 functions. Uses a 16-byte nibble LUT, improving performance over previous unrolled-loop iterative methods, and also producing smaller code (in case of reflect_32, substantially smaller).