Categorical scatter plots on symlog-scaled axis

[MaozGelbart]

Hi,

On the current dev version (eb2b5a2) and matplotlib 3.7.1, consider the following code that uses stripplot to draw two points on unscaled and symlog-scaled yaxis:

import seaborn as sns
import matplotlib.pyplot as plt

x = [0.1,2]
y = [0.1,5]
fig, axs = plt.subplots(ncols=2)
sns.stripplot(x=x, y=y, ax=axs[0])
axs[0].set_yscale("symlog", base=10, linthresh=1)
axs[1].set_yscale("symlog", base=10, linthresh=1)
sns.stripplot(x=x, y=y, ax=axs[1])
axs[0].set_ylim(0,10**4)
axs[1].set_ylim(0,10**4)
axs[0].set_title("stripplot on unscaled axis")
axs[1].set_title("stripplot on symlog-scaled axis")

The plot on the already-scaled yaxis contain values that were not provided. The plot changes (but still erroneous) if I set linthresh to something different (for example if using the linthresh default of 2).
This also happens with pointplot. It works as expected with log-scaled axis or with pure matplotlib scatter calls. Couldn't reproduce using seaborn 0.12.2.