[BUG]: `histogram2dcontour` uses wrong fill colors for negative values

[emilykl]

First reported in plotly.py: plotly/plotly.py#3799

histogram2dcontour traces use incorrect fill colors when bin values are negative.

Negative bin values can occur when histfunc is set to 'avg', 'min', or 'max', and the input z array contains negative values.

When some bin values are negative, all negative bin values are filled with the same color as 0 on the colorscale, while the outermost fill level is filled with the same color as the lowest number on the colorscale.

Example

Codepen reproduction: https://codepen.io/emilykl-code/pen/PwpPjMb

Code:

Show code

const x = [1, 2, 3, 4, 1, 2, 3, 4];
const y = [1, 1, 1, 1, 2, 2, 2, 2];
// The bug is triggered when histfunc is 'avg', 'min', or 'max',
// and some z-values are negative such that some resulting bin values
// end up negative
// This example uses all-negative z values to show the most extreme
// case, but not all z-values need to be negative to trigger the bug
const z = [-10, -50, -100, -150, -10, -50, -100, -150];

Plotly.newPlot('div1', [{
  type: 'histogram2dcontour',
  x: x,
  y: y,
  z: z,
  histfunc: 'avg',  // or 'min' or 'max'
  contours: {
    coloring: 'fill',
    showlabels: true,
  },
}],
 {
   title: {
     text: 'histogram2d contour plot: wrong fill colors!',
     subtitle: {
       text: 'Compare contour labels on plot to colorscale'
     }
 }
});

Screenshot:

Expected behavior

Contour fills should match fill colors indicated by colorscale