ENH: resample - upsampling on level

[jreback]

example from #7895

In [39]: d1 = Index(pd.date_range('1/1/2012', '3/1/2012', freq='MS'), name='date')
    ...: df = pd.DataFrame({'A': [1, 2, 3], 'B': [4, 5, 6]}, index=d1)
    ...: df.columns.names=['id']
    ...: s = df.unstack()
    ...: s.name = 'value'
    ...: 

In [40]: s
Out[40]: 
id  date      
A   2012-01-01    1
    2012-02-01    2
    2012-03-01    3
B   2012-01-01    4
    2012-02-01    5
    2012-03-01    6
Name: value, dtype: int64

In [41]: s.resample('D', level='date').asfreq()
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-41-810faaa64080> in <module>()
----> 1 s.resample('D', level='date').asfreq()

/Users/jreback/pandas/pandas/core/resample.py in asfreq(self, fill_value)
    518         DataFrame.asfreq
    519         """
--> 520         return self._upsample('asfreq', fill_value=fill_value)
    521 
    522     def std(self, ddof=1, *args, **kwargs):

/Users/jreback/pandas/pandas/core/resample.py in _upsample(self, method, limit, fill_value)
    737             raise AssertionError('axis must be 0')
    738         if self._from_selection:
--> 739             raise ValueError("Upsampling from level= or on= selection"
    740                              " is not supported, use .set_index(...)"
    741                              " to explicitly set index to"

ValueError: Upsampling from level= or on= selection is not supported, use .set_index(...) to explicitly set index to datetime-like

This is like #7895 in that we are reindexing on a level (generated via resample). So it equivalent in this case.

This should work.