extend interval to dot and text marks

[Fil]

closes #611

(not related to #597 which requires a very different kind of interval)

[mbostock]

Thanks for incorporating my suggested changes.

I’d love to have a better motivating example of how this should be used. In the case of plotting athletes by birthday month, I think I would probably recommend this instead:

Plot.plot({
  marginLeft: 60,
  y: {
    reverse: true
  },
  marks: [
    Plot.rectX(athletes, Plot.binY({x: "count"}, {y: d => d.date_of_birth, thresholds: d3.utcMonth})),
    Plot.text(athletes, Plot.binY({x: "count", text: "count"}, {y: d => d.date_of_birth, thresholds: d3.utcMonth, dx: 4, textAnchor: "start"})),
    Plot.ruleX([0])
  ]
})

Alternatively, if I wanted an ordinal y-scale using the group transform:

Plot.plot({
  y: {
    tickFormat: Plot.formatMonth()
  },
  marks: [
    Plot.barX(athletes, Plot.groupY({x: "count"}, {y: d => d.date_of_birth.getUTCMonth()})),
    Plot.text(athletes, Plot.groupY({x: "count", text: "count"}, {y: d => d.date_of_birth.getUTCMonth(), dx: 4, textAnchor: "start"})),
    Plot.ruleX([0])
  ]
})

I can reduce your example to this, which produces identical output to the first:

Plot.plot({
  marginLeft: 60,
  y: {
    reverse: true
  },
  marks: [
    Plot.rectX(athletes, Plot.groupY({x: "count"}, {y: d => d3.utcMonth.floor(d.date_of_birth), interval: d3.utcMonth})),
    Plot.textX(athletes, Plot.groupY({x: "count", text: "count"}, {y: d => d3.utcMonth.floor(d.date_of_birth), interval: d3.utcMonth, dx: 4, textAnchor: "start"}))
  ]
})

But, it still seems more complicated since you need to apply the month interval twice, whereas with the bin transform, you only need to specify it once, and the bin transform already computes the midpoint. Though… I guess that alone is a reason for the interval transform to do the same, for symmetry.

[mbostock]

Also, and apologies for not realizing this sooner, we should find a way to reuse the existing mid helper (used by the bin transform) for the interval transform so that the two behave consistently:

plot/src/mark.js

Lines 277 to 292 in 72b2adb

	// Assuming that both x1 and x2 and lazy channels (per above), this derives a
	// new a channel that’s the average of the two, and which inherits the channel
	// label (if any). Both input channels are assumed to be quantitative. If either
	// channel is temporal, the returned channel is also temporal.
	export function mid(x1, x2) {
	return {
	transform(data) {
	const X1 = x1.transform(data);
	const X2 = x2.transform(data);
	return isTemporal(X1) || isTemporal(X2)
	? Array.from(X1, (_, i) => new Date((+X1[i] + +X2[i]) / 2))
	: Float64Array.from(X1, (_, i) => (+X1[i] + +X2[i]) / 2);
	},
	label: x1.label
	};
	}

[mbostock]

This came up again here: https://talk.observablehq.com/t/text-marks-for-the-sum-of-last-n-days-when-using-plot-windowy/6627/2