I've been working on a 4-CPU Arduino for the past couple of months. Photo below is of a recent prototype. It's still pretty rough (no DC jack, USB is still on a breakout, and the pinouts are PCB layout convenient) but you get the idea.
The TX/RX from the USB is switched from CPU to CPU using a DP4T switch. The RES and Power bus lines are tied together as well as the I2C pins (although not totally necessary, was useful in my application).
If I ever wanted to respin this I would probably make on of the CPU pinouts up to date with the Decimilia so shields could be used, and switch most components for SMT for a tigheter layout (right now it's a MONSTER).
Let me know what you think 
[/img]