Implement AlgorithmM

[9il]

• Algorithm M can be found at https://www.researchgate.net/publication/2295884_How_to_Read_Floating_Point_Numbers_Accurately. It is the simplest one. We don't need any other features from this paper or any other algorithm. Exactly AlgorithmM. Don't need to read the whole paper.
• API should be the same as mir.bignum.low_level_view: algoR, @safe @nogc pure nothrow
• Required low-level @nogc pure nothrow arithmetic operations should be added. Don't need to use operator overloading. Standalone functions would work here better.
• The input can be considered valid.

This algorithm is a simple way to add precision parsing for subnormal numbers. Mir still can't do it well. No other D library can do it as well.