ENH: optional dependence on FFTW for faster bandpass

[danielballan]

closes #13

without fftw:

bandpass single pre-loaded image
10 loops, best of 3: 85 ms per loop

batch locate 10 frames
1 loops, best of 3: 11.7 s per loop

with fftw:

bandpass single pre-loaded image
10 loops, best of 3: 60.5 ms per loop

batch locate 10 frames
1 loops, best of 3: 11 s per loop

# Need to run in IPython
import mr
import os

benchmarking code:

path, _ = os.path.split(mr.__file__)
v = mr.Video(os.path.join(path, 'tests/water/bulk-water.mov'))
im = v[0]
mr.bandpass(im, 1, 10)  # slow planning run

print("bandpass single pre-loaded image")
get_ipython().magic(u'timeit mr.bandpass(im, 1, 10)')

print("batch locate frames")
get_ipython().magic(u'timeit mr.batch(v[:10], 9, 1000)')