Move to a decimal floating point library

[louisrubet]

This will help

• getting rid of strange representation problems like Round problem a on very simple operation (divide by 10, substract) #279 ,
• this can also help automatically manage the numbers length in digits (to be defined).

Requirements are: arbitrary precision, conversion to string (several modes possible is a must), transcendent math functions (trig, logs).

Here is a selection of some libs:

lib	license	arb. precision	->string	transcendent
Intel DFP	(c) 2018, Intel	no	yes	yes
libdfp	GPL 2	no	yes	partial (no trig)
libBF	MIT	yes	yes	yes
mpdecimal	simplified BSD	yes	yes	partial (no trig)
IBM decnumber	ICU	no	yes	no
libdfloat	Michael Warren Free Software License	no	yes	no
MathCW	closed	no	?	?

Please see

• https://speleotrove.com/decimal/ as an entry point
• https://discourse.llvm.org/t/rfc-decimal-floating-point-support-iso-iec-ts-18661-2-and-c23/62152/10

Choice goes to libBF from Vincent Bellard.