scale_discrete(drop=True) has no effect (it incorrectly includes unused categories)

[jdanbrown]

from pandas.api.types import CategoricalDtype

class_cat = CategoricalDtype(
    ordered=True,
    categories=['foo', 'bar'] + list(mpg['class'].unique()),
)

Unused categories are incorrectly included with drop=True (the default):

(mpg
    .astype({'class': 'str'})
    .astype({'class': class_cat})
    .pipe(ggplot)
    + geom_bar(aes(x='class'))
    + scale_x_discrete(drop=True)
)

Same as drop=False:

(mpg
    .astype({'class': 'str'})
    .astype({'class': class_cat})
    .pipe(ggplot)
    + geom_bar(aes(x='class'))
    + scale_x_discrete(drop=False)
)

[jdanbrown]

Looks great, thanks!

(mpg
    .astype({'class': 'str'})
    .astype({'class': class_cat})
    .pipe(ggplot)
    + geom_bar(aes(x='class'))
)

(mpg
    .astype({'class': 'str'})
    .astype({'class': class_cat})
    .pipe(ggplot)
    + geom_bar(aes(x='class'))
    + scale_x_discrete(drop=True)
)

(mpg
    .astype({'class': 'str'})
    .astype({'class': class_cat})
    .pipe(ggplot)
    + geom_bar(aes(x='class'))
    + scale_x_discrete(drop=False)
)