position_dodge not working with categorical y and continous x variables

I illustrate the problem with the dots dataset:

import pandas as pd
from plotnine import *

dots = pd.read_csv('https://raw.githubusercontent.com/mwaskom/seaborn-data/master/dots.csv')

Now, I plot with categorical x and continuous y variable:

(
    ggplot(dots, aes(y='coherence', x='align', color='choice'))
    + geom_point(position = position_dodge(width=0.8))
)

This worked fine. Now, I swap the x- and y-axis:

(
    ggplot(dots, aes(x='coherence', y='align', color='choice'))
    + geom_point(position = position_dodge(width=0.8))
)

Now, position_dodge did not work anymore.

ggplot2 gives me what I would expect (I execute the following in a Jupyter notebook)

%load_ext rpy2.ipython

%%R -i dots
library(ggplot2)
(
    ggplot(dots, aes(x=coherence, y=align, color=choice))
    + geom_point(position = position_dodge(width=0.8))
)

Software versions:

import session_info
session_info.show()

-----
pandas              2.2.2
plotnine            0.13.6
session_info        1.0.0
-----

Click to view modules imported as dependencies

-----
IPython             8.26.0
jupyter_client      8.6.2
jupyter_core        5.7.2
-----
Python 3.12.4 | packaged by conda-forge | (main, Jun 17 2024, 10:23:07) [GCC 12.3.0]
Linux-3.10.0-1062.18.1.el7.x86_64-x86_64-with-glibc2.17
-----
Session information updated at 2024-07-15 16:29

Start by running the two plotnine examples from the issue with the dots dataset, comparing position_dodge for categorical x versus categorical y. Trace the position_dodge and geom_point entry points to find why continuous x with categorical y is not separated, then add or update a regression test showing the expected dodged output.