Thanks both. I've incorporated your suggestions and will merge once the checks pass.

One quick example: Based on our discussion in #243, the default for grouped densities is now to use individual bandwidths (also consistent with ggplot2). But users can override and select a joint bandwidth using either the full dataset or one computed as the observation-weighted mean.

tinyplot(~Sepal.Length | Species, data = iris, type = "density") # "none" (default)
tinyplot_add(type = type_density(joint.bw = "full"), lty = 2)    # full dataset
tinyplot_add(type = type_density(joint.bw = "owm"), lty = 3)     # obs-weighted mean
legend("topright", c("None", "Full", "OWM"), lty = 1:3, title = "Joint BW")