Issue with geom_contour_filled when the z values are too close together

[ghost]

There seems to be a bug in the function geom_contour_filled where it fails in some simple cases.

library(tidyverse)

#Generate some data
data = crossing(x = 1:10, y = 1:10) %>% 
  mutate(z = (x+y)/4)

#Works great
ggplot(data) + geom_contour_filled(aes(x, y, z = z))

#Make a minor tweak and it fails
data2 = data %>% mutate(z = z + 100)
ggplot(data2) + geom_contour_filled(aes(x, y, z = z))
#> Warning: Computation failed in `stat_contour_filled()`:
#> factor level [4] is duplicated

Looking into it, I believe the problem is with the legend names (more specifically the function pretty_isoband_levels). If the z values are close, you can get different levels with the same names because it only uses 3 digits (eg. "(101, 101]" and again "(101, 101]").

^{Created on 2022-06-14 by the reprex package (v2.0.1)}