Corrects deadloack between the CoreAudio AudioDeviceStop function and the stream processing callback.

Adds stream capture and render via CoreAudio for MacOSX. This is a first version which needs more testing. The default audio system on MacOSX is still PortAudio.

Modifies the targets 'ffmpeg', 'libvpx', and 'video4linux2' to build the libraries for FreeBSD. Tzanetos Balitsaris

file in order to produce the libjnpulseaudio.so and libjnopus.so shared
libraries for FreeBSD under FreeBSD.

problems were observed when they were used (and we don't yet support stereo
decoding anyway).

opus to be advertised with 2 channels)

(regardless of their associated MediaStream's direction) to pass through
RTPTranslatorImpl

forwards packets. That is, makes it ignore packets received on MediaStream-s
which do not allow receiving, and doesn't send packets on MediaStream-s which
don't allow sending.

assuming a default device direction of SENDRECV.

Addresses issues with hotplugging and playback device selections during calls in the Windows Audio Session API (WASAPI) integration when acoustic echo cancellation (AEC) is enabled.

Handles hotplugging and playback device selections during calls in the Windows Audio Session API (WASAPI) integration when acoustic echo cancellation (AEC) is enabled.

Implements acoustic echo cancellation (AEC) in the integration of Windows Audio Session API (WASAPI) using the Voice Capture DSP in source (as opposed to filter) mode.

Fixes issues with the Renderer implementation using Windows Audio Session API (WASAPI) related to the playback and/or notify device selections being set to 'none'.

Allows opening the CaptureDevice implementation using Windows Audio Session API (WASAPI) when the playback device selection is set to 'none'.