BUG: resample().asfreq() looses end period with TimedeltaIndex

[jorisvandenbossche]

In [16]: df=pd.DataFrame(data=[1,3], index=[dt.timedelta(), dt.timedelta(minutes=3)])
In [17]: df
Out[17]:
          0
00:00:00  1
00:03:00  3

In [18]: df.resample('1T').mean()
Out[18]:
            0
00:00:00  1.0
00:01:00  NaN
00:02:00  NaN
00:03:00  3.0

In [19]: df.resample('1T').asfreq()
Out[19]:
            0
00:00:00  1.0
00:01:00  NaN
00:02:00  NaN

The two examples should give the same result.
Tried it with a DatetimeIndex, and there I don't see this problem.

cc @jreback

[jorisvandenbossche]

Simple test with DatetimeIndex does work:

In [28]: df.index = df.index + pd.Timestamp('2016-01-01')

In [29]: df
Out[29]:
                     0
2016-01-01 00:00:00  1
2016-01-01 00:03:00  3

In [30]: df.resample('1T').asfreq()
Out[30]:
                       0
2016-01-01 00:00:00  1.0
2016-01-01 00:01:00  NaN
2016-01-01 00:02:00  NaN
2016-01-01 00:03:00  3.0

[jreback]

hmm, ok, logic *should8 be the same, but prob not well tested.