Advantages:
* No interpolation needed for creating rougher meshes, lower quality consists in just removing 1 of each 2 lines (3 of each 4 vertices).
* No interpolation needed for morph targets, just apply the morph to the remaining points and skip the rest of them.
* The mesh can be implemented on MakeHuman itself, and have a control to set the mesh quality. Also, it could be configured to use a lower poly mesh when morphing/moving the character, saving CPU resources and improving performance.
Disadvantages:
* The current mesh is complex, some parts would be hard to convert to big quads.
* There has already been a hard work on de-triangulating the mesh, I don't think you'll be up to implement another new mesh. Nevertheless, maybe it's a good idea for exporting models.
* I have never used Blender (well, a few times, but just to see how it looks like), so probably nothing that I've said here makes any sense.