Right but in reality you can't actually hit "always having exactly one thing to process" without maxing out your queue. It's actually impossible! Because they are actually the same thing. "Always having something to process" means no idle time - which means your queue is forever long :) Aiming for that ideal is not a good idea because it's misguided (for lack of a better term).
Come on, it's an ideal; it doesn't have to be realistic.
[Edit] Just read the Slimmons article. His argument is specifically about a queue served by a server that is under stress. His assertion is:
"As you approach maximum throughput, average queue size – and therefore average wait time – approaches infinity."
He's quite right (and yes, it's a bit counter-intuitive). But it's orthogonal to the matter of the ideal queue length.
Thanks for the link.