Fix GroupBy.filter to follow complex group keys.

[ueshin]

Fixing GroupBy.filter to follow complex group keys.

>>> kdf = ks.DataFrame({"a": [1, 2, 3, 4, 5, 6], "b": [1, 1, 2, 3, 5, 8], "c": [1, 4, 9, 16, 25, 36]})
>>> kdf.groupby(kdf["b"] // 5).filter(lambda x: any(x.a == 2))
   a  b  c
0  1  1  1
1  2  1  4

This should be:

>>> pdf.groupby(pdf["b"] // 5).filter(lambda x: any(x.a == 2))
   a  b   c
0  1  1   1
1  2  1   4
2  3  2   9
3  4  3  16

[codecov-io]

Codecov Report

Merging #1471 into master will decrease coverage by 0.04%.
The diff coverage is 66.66%.

@@            Coverage Diff             @@
##           master    #1471      +/-   ##
==========================================
- Coverage   93.74%   93.70%   -0.05%     
==========================================
  Files          36       36              
  Lines        8409     8415       +6     
==========================================
+ Hits         7883     7885       +2     
- Misses        526      530       +4     

Impacted Files	Coverage Δ
databricks/koalas/groupby.py	88.40% <66.66%> (-0.23%)	⬇️
databricks/koalas/frame.py	95.38% <0.00%> (-0.15%)	⬇️
databricks/koalas/generic.py	97.02% <0.00%> (+0.37%)	⬆️

---

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 2e68e03...92a3c5a. Read the comment docs.

[HyukjinKwon]

LGTM except https://github.com/databricks/koalas/pull/1474/files#r422865501

[ueshin]

Thanks! I'd merge this now. I'll address the comment in the following PRs.