Add option to jitter outliers in a boxplot

I would like to be able to add a small amount of jittering to outliers in a boxplot or alternatively stack the points to avoid having them overlap.

Here is an example of where points in a boxplot overlap:

library(ggplot2)
library(dplyr)

# outliers are overlapping
ggplot(mpg, aes(drv, cty)) +
  geom_boxplot()

To add jittering to these outliers, we currently have to result to the following hack, by creating a separate dataset of outliers and plotting them using geom_jitter() manually.

# adding jittering to outliers is a bit of work
outliers <- 
  mpg %>%
  group_by(drv) %>%
  filter(cty > quantile(cty, 0.75) + 1.5 * IQR(cty) | 
           cty < quantile(cty, 0.25) - 1.5 * IQR(cty))

ggplot(mpg, aes(drv, cty)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(height = 0, width = 0.1, data = outliers)

I understand that the position argument in geom_boxplot() is already "occupied", so maybe the simplest solution would probably to just add a new argument outlier.jitter = c(0, 0) (for x and y coordinate jittering respectively).

An even better solution would of course be to incorporate the beeswarm algorithm from ggbeeswarm:

library(ggbeeswarm)

ggplot(mpg, aes(drv, cty)) +
  geom_boxplot(outlier.shape = NA) +
  geom_beeswarm(data = outliers)

^{Created on 2021-05-17 by the reprex package (v2.0.0)}

The issue names no repository files or tests; start by tracing geom_boxplot() and its existing position argument, using the provided geom_jitter() workaround as the behavioral reference. Determine whether the requested behavior should be jittering, stacking, or both, and consider how the option would work for boxplot outliers without a separate dataset.