There's still no control over the channels themselves, but the basic interleaved buffer support is now coded.
There's also new support for buffer size definition in seconds.

Made use of queue support for SDL avoiding the callback approach which greatly simplifies the code and increases the APU accuracy.

It's stil glitchy and noisy, need to check the problems.

Still requires work to be able to be able to emulate APU together with the CPU instead of inverting the control.

This a very work in progress situation that requires stability.
The most important part is the fact that it's possible for an external entity (SDL) to call an audio tick and get a volume.