Heh, I'm not so sure I buy into this definition. This just identifies the state of software configuration, not the cabinet overall.
Says Mr. Hardware Vendor. 
Oookayeeeeee......

But seriously, you could put 4 buttons and a joystick into cutouts in a cardboard box, draw labels on it with a Sharpie marker and a stranger could operate everything without your help....if the software was configured properly. That's a pretty low bar for a "finished" cabinet.
That's why I see it as something that is personal to the builder. Some folks are building for themselves, so the requirement of "on location survival" or "brain-dead usability" doesn't apply at all.
You decide when it is finished, so a project could literally go from "unfinished" to "finished" without ever changing a single part of the project.
IOW, it's a question that can't really have an answer. Now if it were along the lines of "how many folks buy a bunch of parts, get frustrated / bored and never end up building a usable cab (definition: one you can actually play games on)" then we might have something to work with.
RandyT
*edit*
Thought I'd add my take on this in the context of the OP's friend's situation. It's common, but not as common as you might suspect. Arcade parts are a bit costly, so it's unlikely that a failed project remains in a box forever. Eventually, they will get sold by a 3rd party as "never used". A failed project that got a little further ends up on eBay or Craig's list, and we all know what these tend to look like, so spotting one isn't hard. But compared to the numbers of folks buying parts, these occurrences are pretty small. So either the parts are being hoarded in someone's closet without being used, or the "fizzle-out" rate is pretty small. I personally think it's the latter.