Knowledge and Experience, and 100 other users on this forum.
If a mechanic tells you to do something with your car, like a recommendation for oil, or gas, that comes from extensive knowledge and testing. They don't have a "book" to refer to you for referrence. Experience teaches us certain things, and obviously since you are having problems, you sought help at VM Ware, they told you one thing, and you are still having problems because it was easier to simply throw hardware at the problem, rather than actually FIX the problem.
So I can't give you a link. You can search and find many many posts on this subject, why adding more than 1 vCPU to a VM doesn't result in more performance, and many users have reported it actually HURTS performance. So you can search this forum for answers on the subject and find many opinions, but for the most part the majority of users of this community forum are experienced ESX techs with more than 2 or 3 years working with this stuff. So can choose to accept it, but given the fact the info that VM rep gave didn't help you, should tell you something.
Does this mean the 100% utilization within Windows is statistically meaningless?
Yes, largely. It's a point of referrence. It's not absolute.
Seems strange to go from a physical platform with low to normal utilization to a virtual platform running reportedly high utilization
Well physical hardware has it's own drivers, also these OS aren't designed for VM's they just work. The OS wants control of the hardware, the hardware is derived from a software emulated mode that appears as hardware, but it doesn't work like physical hardware. So sometimes the CPU will peg the meter, but if you look beneath the hood in ESX you will see what Windows interprets as 100% CPU may not actually BE 100% CPU core, it's relative performance of what the CPU is set at.
)