Cubic Bézier easing?

[mbostock]

See mbostock/d3#1982. The only problem is that it’s kind of a pain to evaluate a cubic Bézier for y given x (rather than t), so it probably requires a numerical solution.

[Fil]

I've implemented a new function here https://observablehq.com/d/e7d4fef33e304f2b Compared to a reference implementation: we get the same speed, and, unless I'm mistaken, our results are more accurate for the limiting cases of x1=0 and y1=1.