Add a new backend to read audio file and make user able to specify a start time to find a loop point

[NaleRaphael]

mpg123 might not be easily installed on Windows (see also issue #5 ), so that I made some changes in the code base to make user able to switch the backend from mpg123 to audioread (which should be much easier to be used).

And users can choose to install this package with mpg123 or audioread while installing this with pip. (use the command below)

$ pip install git+https://github.com/NolanNicholson/Looper.git --global-option='--backend_audioread'

---

Besides, user can specify a start time to find a loop point from the command line interface by the following command:

$ python3 -m looper track.mp3 --start_time [TIME_IN_SECOND]

This might be a solution for issue #1 .

[ebeers-png]

The changes @NaleRaphael made worked for me! I noticed that although the looping points are seamless, when the music jumps back to the start of the loop there's always short break in the sound. I haven't taken a look at it yet myself, but is there anything you'd suggest to speed up the jump time between the end and the start of the loop? (Not sure if this is a problem for the unforked code, since I wasn't able to use Looper with mpg123.)