inner='box' in violinplot shows too large quartiles and whiskers

[higher-bridge]

When making violinplots, passing inner='box' uses mpl's Axes.plot() to draw the inner boxplot.
However, the default in mpl is solid_capstyle='projecting', which extends the line too far out, and thus exaggerates both the boxplot's quartiles and the whiskers.
This becomes increasingly exaggerated with a larger linewidth, because 'projecting' uses linewidth/2 (https://matplotlib.org/stable/api/_enums_api.html#matplotlib._enums.CapStyle).

Showing incorrect data ranges may be problematic for e.g., scientific publications.
For more accurate plotting, I therefore suggest that seaborn sets the default for inner_kws to solid_capstyle='butt'. This limits the line to its actual endpoints.

I don't know whether this issue may extend to other types of plots as well.

Below is a minimal example which shows both the default and desired behaviours. I am using seaborn 0.13.2.
I set box_width to 12 to make it a bit more obvious, but even at default box_width it happens.

import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns

# List of 0-100
y = np.linspace(0, 100, 100)
avg, q1, q3 = np.mean(y), np.quantile(y, 1/4), np.quantile(y, 3/4)

# To dataframe
df = pd.DataFrame({'y': y})

# Plot default behaviour
plt.figure()
sns.violinplot(data=df, y='y', cut=0, fill=False, inner='box', inner_kws={'box_width': 12})
plt.axhline(avg, c='black', zorder=-100)
plt.axhline(q1, c='black', zorder=-100)
plt.axhline(q3, c='black', zorder=-100)
plt.yticks([q1, avg, q3], ['q1', 'avg', 'q3'])
plt.title('Default behaviour')
plt.show()

# Plot desired behaviour (inner_kws={'solid_capstyle': 'butt'})
plt.figure()
sns.violinplot(data=df, y='y', cut=0, fill=False, inner='box', inner_kws={'box_width': 12, 'solid_capstyle': 'butt'})
plt.axhline(avg, c='black', zorder=-100)
plt.axhline(q1, c='black', zorder=-100)
plt.axhline(q3, c='black', zorder=-100)
plt.yticks([q1, avg, q3], ['q1', 'avg', 'q3'])
plt.title('Desired behaviour (solid_capstyle = "butt")')
plt.show()