The only thing that would have to be considered is that this style also SCREAMS for more animation phases than what we were used to back then - And more animation = more time = more money. Especially character animation would take a much longer time than the typical 16bit 2d stuff. That's why going with 3d and post processing the rendered sprites would make a hell of a lot of sense. Still, that guy said he often needed 30 hours for each of these paintings - 30 hours for one layout! Of course you could reuse a bit of stuff, but that's still not very efficient.
Still, with todays tools and a good pipeline / workflow I don't see any reason why this stuff shouldn't be possible. The question is rather whether or not it's feasible for what one's going to do.