- expose ReferenceError
- remove duplicate fg/bg chans

channels are now dynamically allocated and can be garbage
collected. note that channels do NOT live on the mp heap
so their memory footprint won't trigger a gc, we rely on
pyst3m just doing that on its own every now and then.

- sequencer value mode fixes
- better rms

- system volume per channel
- rms tracking per channel

still missing a few bits here and there but we need a break from
dry exercise and wanna do some pending app update field work instead :P
also it's good to debug a bit in situ too

includes work of dos

/!\ this breaks radspa plugin api

sorry if you were making your own plugins, but it's not hard to port
things, promise. to our knowledge nobody except for us is doing this
anyways. this is a major performance buff compared to 0.2.0 which was
kind of a rushjob and not tested under usual conditions.

in practice, this means you can run gay drums and shoegaze at the same
time again as was possible before 0.2.0, so that's neat.

bl00mbox and radspa have grown up. It is time for them to
gain independence in their very own repositories.

Check out gitlab.com/moon2embeddedaudio for future updates
or contributing.

We will continue to push updates to the flow3r project and
add hardware-specific features.

We see no reason to submodule it right away, but if it is
preferable for the flow3r firmware repo we're open to it.

Thanks for all the support and kind words, we'll try our
best to polish all the rough spots and add a ton of plugins
and features :D!