SpatialAudio: fix QResonanceAudioPlayer to mix additively
QRtAudioEngine zero-fills its output buffer once per audio callback
and invokes play() on every active voice against that same buffer, so
voices are required to mix additively. QResonanceAudioPlayer::play()
instead overwrote. This was latent rather than audible,
since exactly one QResonanceAudioPlayer voice is ever registered per
QRtAudioEngine, but broke the interface's contract.
Also documents the additive-mixing constraint on
QRtAudioEngineVoice::play().
Pick-to: 6.12
Change-Id: I14bdcc72c7ffa6637287e42053cc2ac4608ba988
Reviewed-by: Artem Dyomin <artem.dyomin@qt.io>