Table.remove() converts columns into lists

[davidwagner]

When using Table.remove(), the type of the columns in a table become lists. They should remain arrays. See attached notebook for an example.

TableRemoveOddity.zip

Sample code to reproduce:

>>> newTable = Table().with_columns(
        "Teams", make_array('Aces', 'Bears', 'Cats', 'Dogs', 'Eagles', 'Ferrets'),
        "Scores", make_array(93, 87, 92, 75, 96, 68)
    )
>>> newTable.remove(0)
>>> newTable.column('Scores')
[87, 92, 75, 96, 68]
>>> type(newTable.column('Scores'))
list

Looking at the implementation of .remove() (https://github.com/data-8/datascience/blob/master/datascience/tables.py#L1118), we can immediately see the problem:

for col in self._columns:
            self._columns[col] = [elem for i, elem in enumerate(self[col]) if i not in rows_remove]

It creates a list, not an array. I believe this should be fixed to create an array.

(Seeing as .remove() is also side-effecting and we are trying to promote non-side-effecting methods where appropriate, we could consider deprecating .remove() anyway and pointing people to .exclude().)

Thanks to Dan Kaiser for this bug report.

[abhishict]

I've fixed this issue in the above pull request, could you please review and approve. Thanks!