Add unique function to SeriesGroupBy.

[ueshin]

Adding unique function to SeriesGroupBy.

E.g.,

>>> df = ks.DataFrame({'a': [1, 1, 1, 2, 2, 2, 3, 3, 3],
...                    'b': [1, 2, 2, 2, 3, 3, 3, 4, 4]}, columns=['a', 'b'])
>>> df.groupby(['a'])['b'].unique().sort_index()  # doctest: +SKIP
a
1    [1, 2]
2    [2, 3]
3    [3, 4]
Name: b, dtype: object

Resolves #875.

[HyukjinKwon]

Oh haha I didn't know it can be simple like this.

[itholic]

cool.

[ueshin]

I was surprised too while I was investigating. haha

[itholic]

Maybe deuplicates -> duplicates ?

[HyukjinKwon]

oops, sorry - I missed this.

[itholic]

I'm gonna include this fixing to #1412 my yet unmerged PR.

[ueshin]

oops, thanks!