Fix RollingGroupby and ExpandingGroupby to handle agg_columns.

[ueshin]

Fix Rolling and Expanding to handle agg_columns.

E.g.,

>>> kdf
   a    b
0  1  4.0
1  2  2.0
2  3  3.0
3  2  1.0
>>> kdf.groupby(kdf.a)[['b']].rolling(2).sum()
       a    b
a
1 0  NaN  NaN
3 2  NaN  NaN
2 1  NaN  NaN
  3  4.0  3.0

This should be:

>>> pdf.groupby(pdf.a)[['b']].rolling(2).sum()
       b
a
1 0  NaN
2 1  NaN
  3  3.0
3 2  NaN

Also, it raises errors for other cases:

>>> kdf.b.groupby(kdf.a).rolling(2).sum()
Traceback (most recent call last):
...
pyspark.sql.utils.AnalysisException: "cannot resolve '`a`' given input columns: ...

>>> kdf.groupby(kdf.a)['b'].rolling(2).sum()
Traceback (most recent call last):
...
pyspark.sql.utils.AnalysisException: "cannot resolve '`a`' given input columns: ...

[codecov-commenter]

Codecov Report

Merging #1546 into master will increase coverage by 0.01%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #1546      +/-   ##
==========================================
+ Coverage   94.18%   94.20%   +0.01%     
==========================================
  Files          38       38              
  Lines        8571     8571              
==========================================
+ Hits         8073     8074       +1     
+ Misses        498      497       -1     

Impacted Files	Coverage Δ
databricks/koalas/groupby.py	90.31% <100.00%> (+0.01%)	⬆️
databricks/koalas/series.py	97.76% <100.00%> (+<0.01%)	⬆️
databricks/koalas/window.py	96.78% <100.00%> (+0.42%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e206de2...c6fdc76. Read the comment docs.

[HyukjinKwon]

I couldn't have time to take a close look but seems fine.

[ueshin]

Thanks! I'd merge this now.