You can't say "Trout crushed Donaldson in wRC+, wOBA, etc" and then list his WAR. WAR takes those things into account. It's double counting.
--
People who are saying that Trout having a higher WAR than Donaldson means he should be the MVP in this case are essentially saying "I believe that WAR is accurate enough to have error bars within 0.2 WAR" (or whatever the difference is), and I don't agree with that. WAR is not accurate down to within 0.2, sorry. Is it the best we have? Yes. Does that mean that we should give it to Trout for a difference that likely amounts to noise? No.