This is the key issue. Detection by antibodies or some other label will be the true discovery. Binding that to nanodiamonds or whatever isn't the bottleneck.
Looking at some details on the technology, I would have to disagree. Nanodiamonds seem to be a much better fluorescent label than what is traditionally conjugated to antibodies. Traditional conjugated fluorophores have high susceptibility to bleaching, so you have to sort them out from natural signals. With nanodiamonds, it appears you can bleach all the natural signals and get a clear signal from the nanodiamond. This is pretty cool.
You missed the point. Detecting the protein or whatever that is deregulated in disease early is the hard part, not the fluorophore or nanoparticle to which it conjugates.
Well we have plenty of biomarker candidates (aka human genome project) and, following them will come plenty of monoclonals, which should be relatively easy to humanize. The bottleneck may be 'bioinformatics', and I don't think bioinformatics is going to be a 'embarassingly parallelizable' problem, certainly not in the sense of "throw more people, computers, and bioinformaticians at the problem" sense.