More details about channel up and down-mixing

Hi Everyone,

After some questions from Robert, I've made an effort to give more detail
in the specification about how channels should be interpreted and how
channels should be up and down-mixed for an AudioNode input.  Please see
the enhanced section here:
https://dvcs.w3.org/hg/audio/raw-file/tip/webaudio/specification.html#UpMix

Or, see my diff to the spec here:
https://dvcs.w3.org/hg/audio/rev/b3a132e8d6c7

One of the questions that we've had, is how to handle various multi-channel
cases such as 7.1 versus discrete 8 channel handling.  I've made an effort
to attack that problem, and have suggested a new AudioNode method, so this
type of problem (and others) can be settable on a per-input basis:

setInputMixingMode(...)

which I've described in the spec.

Robert has also suggested extending the pre-defined speaker layouts to some
missing ones, including 3-channel, and 7.1, etc.
I'm happy to work more on the spec to bring it into that direction.

I hope that those interested can take some time to read through this new
text, and I look forward to the discussion.

The new text in the spec does not (or is not intended to) change any
existing behavior of the nodes for the current default stereo output case,
so wouldn't involve breaking developers.  It's intended to address more
advanced multi-channel use cases, and clarify existing behavior.  I still
have a few places in the spec I need to clean up to reference the new text
to make this more clear and coherent.

Cheers,
Chris