Fix colors on Windows (#749)

[stkb]

Looking through the Stack source code lead me to the hSupportsANSIWithoutEmulation function. Calling this on Windows 10+ returns Just True but also enables the processing of the Ansi control codes as a side effect.

This change also has the added benefit that (on Windows), the ANSI codes aren't present when output is redirected to a text file.

Code is just proof-of concept; feel free to modify as you wish. Haven't tested on *nix but it's working on Windows for me.

[stkb]

Oh I was just rebasing at the same time you were :) Will wait for CI then look at any errors

[f-f]

Left small suggestions but otherwise looks great! 👏

[stkb]

@f-f Ok as shown by the failing tests, this actually produces a change in behavior. For all platforms now, if the output is directed into a file, the color codes are never included, even without the --no-color flag.

Is this change desired? If not, we can instead just call the hSupportsANSIWithoutEmulation function but ignore the output.

[f-f]

Good question. It feels like the new behaviour should be more correct, but I don't have a strong opinion on this.
I guess a good call could be to stay consistent with what the compiler does?

[stkb]

I can't find any definitive recommendation on this, but yeah my feeling too is that logfiles should be free from the color codes.

After investigating, this is indeed what the purs compiler does (as well as respecting TERM=dumb). However unfortunately PSA doesn't respect either, so output from that is still colored. I guess that's an issue for PSA though...

[f-f]

Great then, let's move on with this!

[stkb]

Yeah I'm just working on making all the tests pass

[stkb]

Tests are now passing; I stripped the color codes from the relevant test fixtures.

While we were here I also came across the NO_COLOR spec that is another way to disable colors. It seems to have some critics but was very easy to add support for so I've added that too.

[f-f]

This is perfect now, great work! 👏👏

[garyb]

Nice work @stkb!