This avoid the usage of scientific notation.
For float we must convert them to Decimal to get a precision large enough to
show all coefficient digits.
For computed Decimal (which are thus not rounded) we must normalized them first
to avoid trailing 0.
Closes#14522
Cédric Krier