extract_csv is slow and use only one CPU

[rvalyi]

Use to multiprocessing extract each pdf on a different process (then likely a different processor): https://docs.python.org/2/library/multiprocessing.html