Retain name in PeriodIndex resample

[max-sixty]

• closes BUG: Resample loses PeriodIndex name #12769
• tests added / passed
• passes git diff upstream/master | flake8 --diff
• whatsnew entry

[max-sixty]

I'm not sure what the goal of this was. It made this test I added fail

[jreback]

the kind argument is quite tricky. We really should remove it, but that's a whole other project. The issue is that some of the routines are expecting DatetimeIndex so we give it to them even though we are a PeriodIndexResampelr. This all existed before the changed in 0.18.0, but was much trickier to deal with as it was function based. So should be possible to fix now.

[max-sixty]

Am going to punt this to #12774, and loosen the associated test in this PR

[jreback]

pls don't change stuff just because py charm does
use targeted changes

[max-sixty]

I would remove it if I didn't think it made it better / cleaner. Happy to if you don't think it does...

[jreback]

add the issue number

[jreback]

run git diff master | flake8 --diff to see PEP issues

[max-sixty]

@jreback green

[max-sixty]

Although codecov states that this line isn't tested:

Unless I'm mistaken, I added a test for this + a note that the test needed to be expanded to test for the index, rather than only its length. But that requires another PR

[jreback]

no create the expected Series as it comes back
you can fix it later in the other issue

[jreback]

no I just fixed something on codecov

[jreback]

great. put a pointer to this PR in that issue (and maybe a checkbox), so we don't forget.

[max-sixty]

done

[jreback]

looks good. ping on green.

[max-sixty]

@jreback green, apart from the codecov anomaly

[jreback]

thanks @MaximilianR

I moved the tests to the proper location (and some other ones, not sure why there were there). test_resample.

now, if in test_resample you do this (and do it with _ts as well).

def _simple_pts(start, end, freq='D'): rng = period_range(start, end, freq=freq, name='foo') return Series(np.random.randn(len(rng)), index=rng) 

you will get some failures, could just be some test failures though (e.g. the expected need to change).