Use color vision deficiency friendly colormap as default for Clustergram

[joelostblom]

Inability to distinguish red and green is the most common color deficiecy in the population, affecting nearly 10% of people. Although red-green colormaps were once common in heatmaps, especially for microarray data, the color combinations is increasingly avoided for more accessible color combinations such as blue/teal to red/orange/yellow/brown (Colorbrewer has some more examples, check the "colorblind safe" box).

When plotly/plotly.py#1681 lands, I think px.colors.diverging.RdBu_r could be a good option for diverging values, or a map similar to 'coolwarm' from colorcet. For sequential values, viridis, or maybe single or dual colors (e.g. Orange to Red or Reds) could be good.

Related:
http://geog.uoregon.edu/datagraphics/color_scales.htm
www.kennethmoreland.com/color-maps/ColorMapsExpanded.pdf
https://matplotlib.org/3.1.1/tutorials/colors/colormaps.html