Allow math --scale=max for maximum digits

[floam]

math should have an easier way of showing the maximum digits possible for floating point numbers than increasing -s until you get an error (15 currently, which corresponds to FLT_DIG. This could perhaps go up a little bit if we decided to use long double - I plan to do this to tinyexpr.cpp).

The most straightforward thing would be just letting the user pass -1 for a scale that is limited only by precision.

[floam]

I'm a little annoyed I forgot to report this and do it before fish 3.0.0.

[faho]

Do we want just "-1", or also -2, -3 and so on, which seems kinda useless?

Or do we want --scale=max?

[floam]

Do we want just "-1", or also -2, -3 and so on, which seems kinda useless?

Seems useless to me. The way I started doing it was just the special value of -1: floam@cbb7fcb

Or do we want --scale=max?

That might be a better experience for the user.

[ridiculousfish]

+1 to --scale=max

[faho]

Fixed in ca5b7c0.

[floam]

Cool!