Investing tens of thousands of dollars of engineering and processing time isn't worth it to save seconds for an employee working minimum wage (that does the packing for awkward shapes by hand).

When I was in school Dr Klarner was looking into that exact problem. He had it pretty well dialed in on 2d. Once you move to 3d though it becomes a strange problem as what was once optimal for 2d may no longer be in 3d.

In practical terms though time/speed is usually more important in packing. For example you may have optimal packing but it takes 10 hours to physically pack it out is not great. Optimal load size is usually better if you have fixed size packages that interlock in some known pattern or are willing to allow some slack.