Ad blocker-wise I think the best argument is that internalising the traffic blocking can be made a lot more efficient than putting a javascript interpreter in between.

The counter argument is that ad-blocking is an arms-race and the only reason a subset of traffic blocking methods is going to work at all is because there isn't a big enough incentive yet.

Other than that the arguments in favour of V3 are going to be about the same as for any API update. My guess is a few extra APIs and more granular permissions.