I really think some folks are over-engineering this ... much as the guy who wears both a belt, and suspenders.
The first question to ask is: what are we trying to accomplish? A good, solid ground, with no connections to either introduce resistance or noise?
In that case, you simply can't improve on the ground in the dedicated cable that terminates on the device, in a plastic box.
Are you worried that the ground wire might touch another within the panel before it gets terminated on the bus? My first thought is "so what?' ... after all, the box is metal, and will therefor contain any radio noise thus made. If it makes you feel better, though, wrap the exposed portion in the panel. A penny's worth of tape, and you're done.
You want it to go all the way back to the service? Well, good luck. As Greg has mentioned, IBM -who got all this nonsense started in the first place- has backed off, admitting that their concerns were largely unfounded. (It seems that they were being bedeviled by poor wiring practices in general ... rather than any esoteric grounding theory.)
About the only 'improvement' you can make is to up-size the romex. If the orange jacket makes the customer happy, so much the better.
