[DataFrame] Implement loc, iloc

[simon-mo]

What do these changes do?

This PR implements basic loc and iloc for the dataframe.

Out of five allowed inputs, this PR implements the most commonly used three:

• A single label, e.g. 5 or 'a', (note that 5 is interpreted as a label of the index, and never as an integer position along the index).
• A list or array of labels, e.g. ['a', 'b', 'c'].
• A slice object with labels, e.g. 'a':'f' (note that contrary to usual python slices, both the start and the stop are included!).
• A boolean array.
• A callable function with one argument (the calling Series, DataFrame or Panel) and that returns valid output for indexing (one of the above)

Related issue number

[simon-mo]

This PR depends on commit 634dba4 in PR #1599

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/3976/
Test PASSed.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/3988/
Test PASSed.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/3989/
Test PASSed.

[devin-petersohn]

A couple of minor changes. Looking good!

[devin-petersohn]

I prefer to wrap this in list comprehension:

retrieved_rows_remote = [_deploy_func.remote(retrieve_func, self.df._df[partition], idx_to_lookup, col_list) for partition, idx_to_lookup in lookup_dict.items()]

[devin-petersohn]

Would it be faster to do a self._index.groupby('partition')?

[devin-petersohn]

This is great! Thanks @simon-mo !

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/4002/
Test PASSed.

[devin-petersohn]

Passed on private-travis with unrelated failures. OK to merge. Thanks @simon-mo!