add working example

[jeremyandrews]

Here's a functional example, but I feel in a few places (particularly passing chunks into the process_frame_mut) I may not be doing the most efficient thing. A review and any feedback would be greatly appreciated.

Known issues:

• outputs raw data (I used Audacity to File -> Import -> Raw Data... and confirm it's valid audio)
• the upstream rnnoise_demo example also outputs raw data
• it seems Audrey doesn't currently have write-support, though I assume this could be done directly through its dependencies