Disabling grouping along x-axis in grouped boxplot

Using the group keyword argument in Statsplots.boxplot adjusts the distance between individual boxes so that they cluster with their groups (see example below):

using StatsPlots

x = Vector{String}(undef, 40)

for i = 1:40
    if i % 4 == 0
        x[i] = "E"
    elseif i % 4 == 1
        x[i] = "A"
    elseif i % 4 == 2
        x[i] = "B"
    elseif i % 4 == 3
        x[i] = "C"
    end
end    

is_vowel(x) = x == "A" || x == "E" ? "vowel" : "consonant" 

boxplot(x, rand(40), group=is_vowel.(x))

demo_boxplots

Any idea how to disable this behaviour? In my use-case I have three boxes, but want to colour each according to one of two groups - 2 in one, 1 in the other. This asymmetry in distance between boxes makes the plot look really ugly.

Maybe you could submit an issue in StatsPlots.jl to see if an additional argument could be added to the groupedboxplot recipe, where the spacing is defined here.