[db-api] Performance improvements

[john-bodley]

This PR improves the performance of fetchall and fetchmany by iterating on the result set rather than iterating over the cursor's iterator. The following,

import timeit

from pydruid.db.api import Cursor


def execute(self, operation, parameters=None):
    self._results = iter(range(10000))


Cursor.execute = execute


def test():
    cursor = Cursor(url=None)
    cursor.execute(operation=None)
    cursor.fetchall()


print(timeit.timeit("test()", number=1000, setup="from __main__ import test"))

resulted in an execution time of 6.698s and 0.331s for the current and proposed solutions respectively, i.e., the proposed solution is about 20x faster.

to: @betodealmeida @mistercrunch

[mistercrunch]