We figured it's better to start small and increase the limits as we can, rather than have to back-peddle and decrease them, but for most "demo" applications this limit gets you pretty far. Also, we don't document it (we should and will!) but the CPU can burst, so you're not strictly limited to 0.02 cores, that's just what you would be limited to assuming 100% utilization on our end, which shouldn't occur do to our cluster auto-scaling.