PERF: optimize iloc for unique case

[jreback]

related #4017, #4018

optimize the iloc case (which should be the same for unique and non-unique indicies)

In [1]: df = DataFrame({'A' : [0.1] * 30000000, 'B' : [1] * 30000000})

In [6]: idx = np.array(range(30000)) * 99

In [7]: df2 = DataFrame({'A' : [0.1] * 10000000, 'B' : [1] * 10000000})

In [8]: df2 = concat([df2, df2, df2])

In [9]: %timeit df.iloc[idx]
1000 loops, best of 3: 1.49 ms per loop

In [10]: %timeit df2.iloc[idx]
1000 loops, best of 3: 1.41 ms per loop