Pandas different view when index a panel with multiindex

Code Sample, a copy-pastable example if possible

Problem description

I find a bug when using panel.apply(), and with some digging, I find an interesting indexing problem caused by multiindex in Panel Object. It is demonstrated below:

# 
import pandas as pd
pn1 = pd.Panel(pd.np.random.random(size=(3,4,5)),items=['AAA','BBB','CCC'])
print pn1.loc[:,:,1]
pn2 = pn1.copy()
pn2.major_axis = pd.MultiIndex.from_product([list('AB'),list('ab')])
print pn2.loc[:,:,1]

Output:

        AAA       BBB       CCC
0  0.182982  0.774812  0.894589
1  0.654716  0.695453  0.790078
2  0.340826  0.941283  0.938544
3  0.351851  0.971896  0.362317
            A                   B          
            a         b         a         b
AAA  0.182982  0.654716  0.340826  0.351851
BBB  0.774812  0.695453  0.941283  0.971896
CCC  0.894589  0.790078  0.938544  0.362317

With a multiindex as the major_axis of Panel object, the indexing result is transposed. And because of that, when using pn.apply(func, axis=('items','major')), it give you the wrong results when the major_axis is a multiindex. Is this a bug? if not, then what is the purpose that it is designed like this? Thanks for help.

Comment From: jreback

duplicate of https://github.com/pandas-dev/pandas/issues/6286

Panel is being deprecated, mainly because of the difficulties of indexing :> see https://github.com/pandas-dev/pandas/issues/13563

for multi-dimensional data would recommend a DataFrame with a multi-index or use https://github.com/pydata/xarray