Pydub

- Rename `pydub.AudioSegment().set_gain()` to `pydub.AudioSegment().apply_gain_stereo()` to better reflect it's place in the world (as a counterpart to `pydub.AudioSegment().apply_gain()`)

- Add `pydub.AudioSegment().pan()` which returns a new stereo audio segment panned left/right as specified.

- Add a logger, `"pydub.converter"` which logs the ffmpeg commands being run by pydub.
- Add `pydub.AudioSegment().split_to_mono()` method which returns a list of mono audio segments. One for each channel in the original audio segment.
- Fix a bug in `pydub.silence.detect_silence()` which caused the function to break when a silent audio segment was equal in length to the minimum silence length. It should report a single span of silence covering the whole silent audio segment. Now it does.
- Fix a bug where uncommon wav formats (those not supported by the stdlib wave module) would throw an exception rather than converting to a more common format via ffmpeg/avconv

- Add `pydub.AudioSegment().max_dBFS` which reports the loudness (in dBFS) of the loudest point (i.e., highest amplitude sample) of an audio segment

- Overhaul Documentation
- Improve performance of `pydub.AudioSegment().overlay()`
- Add `pydub.AudioSegment().invert_phase()` which (shocker) inverts the phase of an audio segment
- Fix a type error in `pydub.AudioSegment.get_sample_slice()`

- Add `pydub.generators` module containing simple signal generation functions (white noise, sine, square wave, etc)
- Add a `loops` keyword argument to `pydub.AudioSegment().overlay()` which allows users to specify that the overlaid audio should be repeated (i.e., looped) a certain number of times, or indefinitely