Provide grid object for univariate distributions of columns in a dataframe

[mwaskom]

i.e.

df = sns.load_dataset("penguins").select_dtypes(include=float)
f, axs = plt.subplots(1, df.shape[1], figsize=(df.shape[1] * 3, 3), constrained_layout=True)
for col, ax in zip(df, axs):
    sns.histplot(df[col], ax=ax)

A few options:

• Make this a special case in displot with an internal melt (-.5)
• Make this a special case in PairGrid with an option to make the x or y axis "derived" (-.5)
• Create a new grid object (name??) that does this, along with a figure-level function (+.5)

It should probably support hue, col_wrap, and possibly support putting the data variable on y.