Bug summary

when using matplotlib.pyplot.axis('equal') with a polar plot, it tends to make things completely unreadable. This was discovered after original using a Cartesian plot and changing the one line of code from plt.plot to plt.polar and the plt.axis('equal') was left in. Right now, the clear solution is to not change the axis attributes using plt.axis('equal') when using a polar plot.
However, maybe the axis attributes that would create a messed up plot should be ignored if the plot is a polar plot.

import matplotlib.pyplot as plt
import numpy as np
data_angles=np.arange(0,2.05,.05)*np.pi
data_radii=np.sin(data_angles)
plt.polar(data_angles, data_radii)
plt.axis('equal')

Matplotlib version

• Operating system: CENTOS 7
• Matplotlib version: 3.1.0
• Matplotlib backend: GTK3Agg
• Python version: 3.6