Use absolute indexes that we let wrap around. We can then easily detect
how much we under of overflowed by using serial number arithmetic.
Remove the Areas, we can trivially compute this ourselves, move the
logic in read/write_data.
The proxy/resource ids are generated by the client and so we need to
used them as index in the client resource map instead of making our own
number.
Fix stream disconnect and client-node destroy
Fix gstreamer device provider
Make all sockets non-blocking to avoid errors with bad clients
Advertise the subsystems we monitor and disable the gstreamer monitors.
Implement core properties updates
Make sure we send REMOVE_ID after we are done with the resource.
Remove queue and ringbuffer between nodes. transfer the buffer id
directly between the io areas when possible.
Let only pinos send push or pull requests for now.
Allow polling multiple fds, like how alsa wants it
Remove port_id from events.
