Match audio-device to upstream implementation

Fix sample looping

Add sample xfading

Refactor next sample getter

Add initial wait-for-zero-crossing implementation

Refactor sample data into struct

Fix multichannel audio

Use u8 samples instead of s8

Add multi-channel audio

Prepare for multi-channel behaviour

Add important TODO comment

Add separate stacks for audio vectors

Change envelope implementation to be per-sample

Add stereo volume and sample interpolation

Wire uxn ADSR to audio device

Implement basic ADSR envelope

Add wip audio callback routine for new audio device

Update build system to change roms without make clean

Make FRAME_TIMER a config parameter to be able to speed up evaluation

Add sample player (with caveats)

The UXN ADSR doesn't behave like it would work usually (with sustain
being a level amount after the decay), instead it splits an envelope
into 4 regions that each can last up to 1 second. In this context, the
sustain in UXN is the amount of time (0-1sec) in which the level is kept
to a maximum level (1.0).

This means we can't fully interop with the built-in ADSR on the
Playdate, so I've made the decay behave as a release and the sustain and
release will always be 0.0.

To obtain the correct behaviour it probably be needed the usage of
a custom signal callback to be used as a volume modulator.