The thing is that this thought experiment is interesting, but it is wrong. It is true that it is not logically possible for a system of two objects to fall faster than both of its constituent parts, but that does not mean that it is a priori logically impossible for heavier objects to fall faster than lighter objects - the lighter object would slow down the heavier object somewhat, so the new "combined" object would fall slower than the heavier of the two objects, but it would fall faster than the lighter of them.

In fact, the fact that heavy and light objects fall at the same speed is a profoundly special property of the gravitational interaction. No other fundamental force behaves this way: in an electric field, objects with more charge will accelerate more quickly than objects with less charge (and the same is true for the weak and strong forces).

In fact, the acceleration of an object or particle in a field is proportional to its field-specific charge divided by its inertial mass. The really interesting thing about the gravitational interaction is that the "charge" associated with the gravitational field is exactly equal to the inertial mass of that object, so all objects accelerate at the same rate because of the gravitational field.

This truly special property of the gravitational field had no explanation until Einstein's theory of general relativity, which discarded the idea that the gravitational field is a field at all, and described the motions of objects only in terms of rectilinear movement in a curved space-time (the reason why inertial mass curves space time is still unexplained though - it seems to simply be a property of the universe).