Prediction plots for logistic models: improvements

The current default plot for logistic models is like that:

p <- plot(modelbased::estimate_relation(glm(vs ~ mpg, data = mtcars, family = "binomial")))
p

However, when the outcome variable is not on the 0-1 scale, but is a factor (and thus 1-2 by default), the plot is messed up:

p <- plot(modelbased::estimate_relation(glm(sex ~ body_mass_g, data = palmerpenguins::penguins, family = "binomial")))
p

^{Created on 2021-06-09 by the reprex package (v1.0.0)}

What would be the most elegant solution to plot the line on the outcome's scale?

On a related note, I'm thinking about moving from a pseudo-rug geoms for data points towards maybe something that would look like this:

What's the best way to create something like that? Using ggdists?

Start with the estimate_relation() and plot() examples in the issue, comparing the 0–1 logistic outcome with the factor outcome shown in the linked images. Review the referenced barelysignificant.com example and the question about ggdists; done should define an agreed approach for plotting the line on the outcome scale and replacing the pseudo-rug presentation.