Normalizes the value in the constructor rather than at each operation
Inlines and optimized add, sub, mul, div
@ckormanyos Prior to this PR the benchmarks were showing that decimal32_fast was running in 0.686 of the time that decimal32 was. Now it's down to 0.251.
@ckormanyos Prior to this PR the benchmarks were showing that
decimal32_fast
was running in 0.686 of the time thatdecimal32
was. Now it's down to 0.251.Closes: #619