bivariate displot with kind="hist" ignores common_norm=False if hue is not assigned

[mgab]

Somehow similar to #2377, but for kind='hist' and with common_norm=False being ignored.

If I understand it well, passing common_norm=False to sns.distplot should produce plots with independent "z" scale. While this is the behaviour for kde plots, it seems to be ignored for bivariate histograms unless hue is also specified.

Test case:

x, y = np.random.randn(2, 200)
z = [0] * 150 + [1] * 50
sns.displot(x=x, y=y, col=z, kind="hist", cbar=True, common_norm=False)
sns.displot(x=x, y=y, col=z, hue=z, kind="hist", cbar=True, common_norm=False)

Running seaborn version 0.11.1
(I tried to look for other (open or closed) related issues, hope not duplicating them)

[mgab]

Uhm... it seems to come from line https://github.com/mwaskom/seaborn/blob/master/seaborn/distributions.py#L743

Should it be something like this?

common_color_norm = not set(self.variables) - {"x", "y"} or common_norm

Similar to what it's done here https://github.com/mwaskom/seaborn/blob/master/seaborn/distributions.py#L723

I'm not familiar at all with seaborn's code, so not sure at all I'm not missing something :P
Could prepare a PR if that looks like a fix

[mwaskom]

Hm, yes, seems like the inverse of the problem in #2377