Add ndensity computed group stat to StatBkde2d

[seasmith]

Gives stat_bkde2d() the same behavior as stat_density2d() -- normalizing the density to the max density per group when facetting. Similar to the nlevel computed stat.

Before (density)

library(ggplot2)
library(ggalt) # from seasmith/ggalt branch add-ndensity

ggplot(iris, aes(Sepal.Length, Petal.Length)) +
    stat_bkde2d(aes(fill = stat(density)), geom = "raster", contour = FALSE) +
    facet_wrap(vars(Species))

After (ndensity)

ggplot(iris, aes(Sepal.Length, Petal.Length)) +
    stat_bkde2d(aes(fill = stat(ndensity)), geom = "raster", contour = FALSE) +
    facet_wrap(vars(Species))

[hrbrmstr]

Very welcome addition! Can you also add yourself to the DESCRIPTION?