Update boolean scheme9 scales

[mkfreeman]

The lighter color of the boolean color scales is difficult to see against a white background. This branch updates the default scale. See this notebook to compare results.

[zanarmstrong]

+1 to this feature!

(how light the hue was came up in data viz dev pairing this morning, and @mkfreeman ran with it!)

[mkfreeman]

I see that I need to update the test (2) plot colorSchemesOrdinal:), looking into that now.

[mbostock]

This PR looks good. I’ve also now enabled branch restrictions for this repo to make it explicit that approval is expected before merging. 🙏

[mkfreeman]

Thanks! I wasn’t sure if the review above was explicit approval, but was recently told that the Author of the PR should merge (so when I saw the button enabled, I thought it was on me).

[mbostock]

No, that was a comment. An approval looks like this:

You should merge, but only after it’s approved. (I’ve now made that restriction explicit.)

[zanarmstrong]

Just curious about the motivation for the switch from:
if (n === 2) return [scheme[5][1], scheme[5][3]];
to
if (n === 2) return [scheme[3][1], scheme[3][2]]
?

The most important goal (to me) was to avoid the lightest end of the spectrum, which is now solved for*.

But, sometimes the dark end is pretty dark too. So, when we were testing things, I thought that [5][1] / [5][3] was a nice balance between getting a lot of intensity difference between the two boolean colors while also avoiding the darkest darks and lightest lights.

All in all, I'm super happy with this change and am 100% fine leaving it as if (n === 2) return [scheme[3][1], scheme[3][2]].

That said, I am curious to understand why you prefered [3][1]/[3][2] to [5][1]/[5][3] :). Thank you!

* except the Magma / Viridis / Plasma family which go from dark-to-light by default instead of light-to-dark like "greys", etc.

[mkfreeman]

This was @mbostock's suggestion, but to me it [3][1] - [3][2] feels less arbitrary: [5][1] - [5][3] felt like a guess (I still wasn't sure that the colors were dark enough).

[Fil]

The general hand-wavy argument is that if a scheme is good for 3 colors, then a pair extracted from that scheme should be good for 2 colors?

[zanarmstrong]

sounds good - thanks!