Also not pointed out is co-dependent optimization problems.

A real world example would be heads-up poker. One player has position on the other so the game is asymetric. The actions of the first player effects the optimal play for the second player, and vise-versa.

If you are trying to solve for both players then SA's are going to be very awkward to leverage.

GA's on the other hand lend themselves naturally to this sort of problem.