The problem with that analogy is that the top half of that lunchbox is adding an equal amount of it's own torsional rigidity to your overall ability to twist said item. The hatch on the Mustang would be the equivalent of bolting a piece of cardboard in place of the lunchbox lid and then expecting it to have some significant impact on your ability to twist the box.
I suppose I can conceded that the hatch will offer some sort of resistance, but the effect is so negligible that it is in all practical purposes non-existent. But, we will just have to agree to disagree on this effect because neither of us can produce any numbers to prove it one way or the other.
The weight savings aren't in so bad a spot as you think. The weight of the hatch is quite high up on the car which is bad from a handling perspective. Given the percentage change in front to rear bias from the slightly less than 20 lbs you save from going to fiberglass from steel, I think it is still a net gain. Plus, as I've always said, you can always add ballast down low if you want more weight on the rear. Of course, no one adds weight unless they are trying to meet a minimum weight rule.
One other thing. Fiberglass doesn't rust