Using mdates.ConciseDateFormatter in multiple subplots gives erroneous offset in the axis:
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.dates as mdates
dti = pd.to_datetime(["2016-08-31", "2016-09-30"])
ts0 = pd.DataFrame({"x": [0, 1]}, index=dti)
ts1 = pd.DataFrame({"x": [0, 1]}, index=dti + pd.Timedelta(365, "D"))
fig, axs = plt.subplots(2, 1, sharex=False)
ts0["x"].plot(ax=axs[0], marker="o")
ts1["x"].plot(ax=axs[1], marker="o")
dlocator = mdates.AutoDateLocator(minticks=6, maxticks=9)
axs[0].xaxis.set_major_locator(dlocator)
axs[0].xaxis.set_major_formatter(mdates.ConciseDateFormatter(dlocator))
axs[1].xaxis.set_major_locator(dlocator)
axs[1].xaxis.set_major_formatter(mdates.ConciseDateFormatter(dlocator))
The figure shows the x-axis tick labels for the top plot got messed up when assigning the major locator and formatter to the bottom plot. Is there any workaround to this bug or problem?

