IMO, the card was probably scrapped because it wouldn't be substantially faster than a 480.
I could see this, 460 SLI is about 15-20% faster than a single 480 but comes with a smaller bus and memory size, so for some the 460s are not appealing if they are using a high resolution.
How much would be gained from having the gf104 cores with those few more unlocked shaders is unknown though. There certainly would be a boost.
Also, would it of been worth it ? Look at benches of 460 sli vs a 5970, if they used fully unlocked gf104 cores it would of been certainly been faster, but it would of used more power.
There is a good chance they couldn't of even produced the card and kept it below the 300W PCIe power spec.
And, dual-gpu cards are a product you release to attain the bragging rights crowd and the waterfall effect given to your full lineup of products that having the fastest card on the market gives. There is no point in going for that position just to lose it within a month or two to AMD's upcoming 6 series cards.
If anything, so quickly producing said card and having it rapidly eclipsed by your competition could have a negative effect on opinion of your products as a whole rather than the positive light having the fastest video card gives AMD or nvidia.
I think nvidia is just going to turtle in the performance crown area until they can get their act back together and try to make a release timed to meet one of AMD's that puts them not only ahead in performance but also ahead or on parity in release cycle timelines.
Clearly the 480 did not do much for them even with its performance edge over a 5870. It came too late and did not give a remarkable difference for most buyers over what was already here in the 5870. And only five months later we're now on the cusp of another AMD release neutering whatever performance advantage the 480 had.
They need to pull off another 8800GTX type release before wasting their time on a dual-gpu part, that in the current climate of competition from AMD, will likely not sell well at all.