requirements: support embedded targets

[dvdsk]

See also the discussion here: #9

Few notes from me:

• This need is currently fulfilled by https://github.com/10buttons/awedio
• It has never seriously been requested from Rodio (except maybe this)
• Once hardware get fast enough to support more complex audio operations it can generally run an OS & do audio.

[CorvusPrudens]

Maybe a more actionable goal that's highly correlated would be no_std support. As I mentioned here, no_std support opens the door to all sorts of platform support without requiring too much work from authors.

Significant parts of Bevy already support no_std. In fact, one contributor has got the engine running on a GameBoy. Now, audio engines won't be running on a GameBoy, but I frequently work with microcontrollers that have the capability.

[bushrat011899]

With libm and portable-atomic, I'd actually say these requirements can be relaxed without sacrificing the internal development experience.

I'd say the "hard" requirement should be no_std support (not necessarily full functionality), but leave it as an exercise for the reader how to actually make their weird platform work.

This is a similar approach to how wgpu is going for no_std support; provide both sides of a public API.