I don't see anything like that. In that last image, in the same column "not in top 10" there's the project ranked #3 on the image above, and the project ranked #17.

In this image, two projects mention being upvoted by an "expert", and the two are in the top 10 indeed.

Overall, I find that the evidence fails to convey the point, I may be wrong and the case is strong but right now, I don't think it's convincing that anything shady is happening.

I'm not surprised that the ultimate choice of giving out money is closely guarded, because the very little amount of voting happening makes it very sensible to fraud. (When a photo of an egg can get 20 million likes on twitter, basing a contest on handfuls of votes would seem a bit...naive.)

That is not necessarily a bad thing. If the experts were highly correlated with what the rankings already were, there would be no point in using them. You want as many independent inputs as possible; the more correlated they are, the more redundant they are.