new algorithm(xjb) for double to shortest decimal. #238

hello，

link : https://github.com/xjb714/f2dec_bench

If it doesn't work, please give me feedback.
It seems a little difficult to prove that the algorithm is correct, but I can't find a counterexample.
The algorithm may be correct, and I have tested a large number of data with correct output.
I have tested all the float values and the output results are correct.
If you are interested, you can study my algorithm code, and the function name is xjb_f64_to_dec.
If you know that others are interested in this kind of algorithm, you can also share it with others.

Thanks for reading.