Enhanced voice support: receive, stereo, encryption

[SpaceManiac]

The voice module should support:

• [x] the opus crate instead of internal bindings
• [x] receiving voice
• [x] stereo transmission, in addition to mono
• [x] encrypted voice (non-plain mode)
• [x] multii-server voice

Voice receive improvements to be made (based on suggestions by #q):

• [x] Supply stereo: bool explicitly to AudioReceiver::voice_packet
• [ ] Handle receiving opus packets larger than 20ms (960 samples)
• [ ] Destroy entries in the decoder_map if they haven't been used for some time
• [x] Document the unreliability of speaking_update for anything but the ssrc mapping
• [ ] Deal with UDP connection timeout if nothing is received for a while