This allows a voice/source to start with a delay, more accurately than simply
waiting to call alSourcePlay. The delay is affected by pitch and velocity,
making it useful to simulate distant sounds that take time to be heard.
It's not available as an AL buffer format (yet) since I'm not sure how to
expose it. Internally it seems fine as a separate channel configuration, but
because OpenAL combines the channel configuration and sample type, a flag may
work better there.
This attempts to correct for the differences needed for 2-channel UHJ's shelf
filters given the output shelf filters. It's far from ideal, but better than
nothing.
There's now effectively a 16-channel limit for buffers (as determined by the
number of elements in DeviceBase::mSampleData). Any more than that are
ignored when mixing.
