Public issue https://github.com/mackron/miniaudio/discussions/800

Public issue https://github.com/mackron/miniaudio/issues/779

Use the MA_PTHREAD_REALTTIME_THREAD_PRIORITY define.

This is useful because MA_ASSERT is only defined in the implementation
section of miniaudio.h which can cause issues when people copy/paste
the code and use it in a file that does not have visibility of the
implementation.

Note that there are still more references to implementation-defined
macros, but these have been moved to the public section in the dev-0.12
branch so I'm not bothering to change those just yet.

Public issue https://github.com/mackron/miniaudio/issues/787

Use `#define MA_AAUDIO_MIN_ANDROID_SDK_VERSION 26` if you need to
support SDK version 26.

Define MA_AAUDIO_MIN_ANDROID_SDK_VERSION to specify the minimum
required SDK version for enabling the AAudio backend.

Public issue https://github.com/mackron/miniaudio/issues/781

Audio Worklets do not work with --closure=1 because the callback used
with emscripten_create_wasm_audio_worklet_processor_async never gets
fired which means miniaudio will never be able to escape from it's busy
wait loop.

Public issue https://github.com/mackron/miniaudio/issues/778

Public issue https://github.com/mackron/miniaudio/issues/779

Unlocked notification fires on emscripten upon successful resume of
audio context attached to a device.  This occurs only once per device
and it happens after the browser has received the input event
necessary to begin playing audio on most webpages.  This is due
to autoplay rules.

It is recommended to wait until this event is fired to start a
'main game loop' on the web.

Public issue https://github.com/mackron/miniaudio/issues/773

Public issue https://github.com/mackron/miniaudio/issues/770

This also fixes a possible crash during initialization due to leaving a
thread running after early termination of the initialization routine.

Public issue https://github.com/mackron/miniaudio/issues/759

occured -> occurred
aquired -> acquired
accomodate -> accommodate
seperate -> separate
etc.

When seeking, the seek does not immediately get applied to the data
source. Instead it is delayed. This results in a situation where the
sound can be seeked, but a small window will exist where querying the
seek point will not be consistent with the previous seek request.