dense interval for line and area

[mbostock]

Fixes #597. Fixes #351.

With the default reduce = sum:

With reduce = median:

[mbostock]

In terms of the example in #351 (comment), I don’t think I would use the new interval option introduced by this PR, and would instead add filter = null to generate dense bins in the first bin transform.

Plot.plot({
  y: {
    axis: "right",
  },
  facet: {
    data: penguins,
    y: "species",
    marginLeft: 80,
  },
  marks: [
    Plot.areaY(
      penguins,
      Plot.stackY(
        Plot.binX(
          { y: "count", filter: null },
          { x: "body_mass_g", thresholds: 20, fill: "sex", fillOpacity: 0.2 }
        )
      )
    ),
    Plot.ruleY([0]),
    Plot.lineY(
      penguins,
      Plot.stackY2(
        Plot.binX(
          { y: "count", filter: null },
          { x: "body_mass_g", thresholds: 20, stroke: "sex", marker: "circle" }
        )
      )
    ),
  ],
})

[Fil]

Could we change the default reducer to "first"? "sum" only works with a quantitative y, whereas "first" (or "mode", but it's slower and not that better) would also work with an ordinal y.

[mbostock]

Could we change the default reducer to "first"?

Okay, done. This means you’ll see gaps by default, which is probably a good thing… and I suppose adding reduce = sum isn’t too hard to remember?