Make BEES serial protocol less crash-prone

I'd still love to see the aleph totally resilient to serial-traffic Denial-of-Service attacks - why should communications overload bring down the system? Can we somehow figure out why aleph dies when flooded with serial commands faster than it can process them?

If not possible/worth it to figure that out, a bit of ACK/NAK handshaking in the aleph serial protocol would at least allow the host software to wait for aleph to respond to each message, rather than forcing host software to sleep after each message (as we're doing currently)...