Introduce RTCAudioSession.initializeInput

Add org.webrtc.AutomaticHardwareVideoEncoderFactory to Android SDK

Do not try to have a "perfect" scale factor in webrtc::VideoAdapter

Without this change, webrtc::VideoAdapter tries to have a scale factor
which can be represented in an easy fraction. That behavior requires
the cropped image to be aligned with scale fraction denominator multiplied
with the output alignment requirement, which can be much larger for
relatively large output alignment requirement.

The behavior does not seem to be required by any implementation so just
abandon it to avoid such a large alignment requirement.

Align video resolution by 16 for VideoToolbox hardware H.264 encoder

Update Unity from 2019.2.6f1 to 2019.2.14f1

Allow to use audio units other than Voice-Processing I/O unit

[sketch-ios] Disable output in VoiceProcessingAudioUnit

Do not initialize parameters with audio session if audio disabled

Implement [RTCAudioDeviceModule deliverRecordedData:]

Introduce .NET bindings