Miniaudio

- Fix a bug with UWP detection.

- Automatically switch the internal device when the default device is unplugged. Note that this is still in the
early stages and not all backends handle this the same way. As of this version, this will not detect a default
device switch when changed from the operating system's audio preferences (unless the backend itself handles
this automatically). This is not supported in exclusive mode.
- WASAPI and Core Audio: Add support for stream routing. When the application is using a default device and the
user switches the default device via the operating system's audio preferences, miniaudio will automatically switch
the internal device to the new default. This is not supported in exclusive mode.
- WASAPI: Add support for hardware offloading via IAudioClient2. Only supported on Windows 8 and newer.
- WASAPI: Add support for low-latency shared mode via IAudioClient3. Only supported on Windows 10 and newer.
- Add support for compiling the UWP build as C.
- mal_device_set_recv_callback() and mal_device_set_send_callback() have been deprecated. You must now set this
when the device is initialized with mal_device_init*(). These will be removed in version 0.9.0.

- Add support for specifying the size of a device's buffer in milliseconds. You can still set the buffer size in
frames if that suits you. When bufferSizeInFrames is 0, bufferSizeInMilliseconds will be used. If both are non-0
then bufferSizeInFrames will take priority. If both are set to 0 the default buffer size is used.
- Add support for the audio(4) backend to OpenBSD.
- Fix a bug with the ALSA backend that was causing problems on Raspberry Pi. This significantly improves the
Raspberry Pi experience.
- Fix a bug where an incorrect number of samples is returned from sinc resampling.
- Add support for setting the value to be passed to internal calls to CoInitializeEx().
- WASAPI and WinMM: Stop the device when it is unplugged.

- Add sndio backend for OpenBSD.
- Add audio(4) backend for NetBSD.
- Drop support for the OSS backend on everything except FreeBSD and DragonFly BSD.
- Formats are now native-endian (were previously little-endian).
- Mark some APIs as deprecated:
- mal_src_set_input_sample_rate() and mal_src_set_output_sample_rate() are replaced with mal_src_set_sample_rate().
- mal_dsp_set_input_sample_rate() and mal_dsp_set_output_sample_rate() are replaced with mal_dsp_set_sample_rate().
- Fix a bug when capturing using the WASAPI backend.
- Fix some aliasing issues with resampling, specifically when increasing the sample rate.
- Fix warnings.

- Fix a crackling bug when resampling in capture mode.
- Core Audio: Fix a bug where capture does not work.
- ALSA: Fix a bug where the worker thread can get stuck in an infinite loop.
- PulseAudio: Fix a bug where mal_context_init() succeeds when PulseAudio is unusable.
- JACK: Fix a bug where mal_context_init() succeeds when JACK is unusable.

- Fix a bug on macOS with Core Audio where the internal callback is not called.