I would tend to agree more with a composite than a "singular anatomical system" approach, when it comes to assessing predatory capability. There's no doubt troodontids were *capable* of predation, and a lot of literature still seems to argue that ambi-manual "clapping" a la Ostrom's "path to flight" theory is a useful mechanism to explain these large limbs. Seeing how oviraptorids have the same forelimbs and are now more or less considered almost exclusively herbivores, there's a lot of speculation that forelimb functionality is *relevant*. But it must have been useful, if not for climbing, digging, brooding, nesting, or perhaps even for defense (using claws will inflict a great deal more damage than engorged knuckles or wrists as in screamers and other galloanserans), than for something like ... prehension.
Multivariate approaches tend to have a better picture, but diet is chiefly assessed through principle processing, not necessarily acquisition method, and this seems to lean closer to the dental assessment of Fan et al., which strongly favors dental and cranial anatomy than single-unit systems, and that data suggests something not unlike an herbivore, though I'm cautious about this given that troodontids don't have many other features assessing them as even plant-dominant herbivores other than denticle morphology and isotopes (which, as I would suggest, are more compelling again than single-unit pronouncements).
Could troodontids (in general, not a specific species) be omnivores, catching prey and eating tough-fibrous plant matter alike? There's hardly any conclusive evidence for the lack of this faculative omnivory, so I fear this new paper does little to deflect from that view.
And really, we need to either reassign the holotype of Troodon formosus or please, please: stop calling it "Troodon formosus."
Cheers,