.align() sorts indexes when they differ, does not sort otherwise

[ipozdeev]

Code Sample, a copy-pastable example if possible

# two frames w/similar indexes
df_1 = pd.DataFrame(np.eye(4))
df_2 = pd.DataFrame(np.eye(4)*-1)

# reverse, sort -> indexes are kept reversed
df_1.iloc[::-1].align(df_2.iloc[::-1], axis=0, join="outer")

# one frame has a different index
df_2 = pd.DataFrame(np.eye(4) * -1, index=range(1, 5))

# reverse, sort -> indexes are sorted
df_1.iloc[::-1].align(df_2.iloc[::-1], axis=0, join="outer")

Problem description

Inconsistent behavior, makes it impossible to use structures of the form:

def func(x, arg):
    if arg < 0:
        return func(arg.iloc[::-1])
    return ...

as the order gets switched without user's consent.

Expected Output

indexes are kept in the original order

Output of pd.show_versions()

[paste the output of pd.show_versions() here below this line]
INSTALLED VERSIONS

commit: None
python: 3.5.2.final.0
python-bits: 64
OS: Windows
OS-release: 7
machine: AMD64
processor: Intel64 Family 6 Model 58 Stepping 9, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: None.None
pandas: 0.23.1
pytest: 2.9.2
pip: 9.0.1
setuptools: 23.0.0
Cython: 0.24
numpy: 1.11.3
scipy: 0.18.1
pyarrow: None
xarray: 0.9.3
IPython: 4.2.0
sphinx: 1.3.1
patsy: 0.4.1
dateutil: 2.5.3
pytz: 2016.4
blosc: None
bottleneck: 1.2.1
tables: 3.2.2
numexpr: 2.6.1
feather: None
matplotlib: 1.5.1
openpyxl: 2.3.2
xlrd: 1.0.0
xlwt: 1.1.2
xlsxwriter: 0.9.2
lxml: 3.6.0
bs4: 4.4.1
html5lib: None
sqlalchemy: 1.1.13
pymysql: 0.7.9.None
psycopg2: None
jinja2: 2.8
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

[TomAugspurger]

Haven't looked yet, but I assume that's a result of Index.union sorting when they're not aligned.

In [3]: pd.Index(['b', 'a']) | pd.Index(['b', 'c'])
Out[3]: Index(['a', 'b', 'c'], dtype='object')

so this is blocked by #17839 at least.