1. There is variance in matter distribution. If fusion created or expanded space I would assume there'd be a measurable variance in the wavelength of the cosmic microwave background;
2. The initial phase of the Universe was a massive period of inflation. This was before atoms could exist (and thus fuse) so to be consistent this would have to be a different mechanism;
3. The energy released by fusing light nuclei comes the nuclear binding energy (ie the strong interaction). So the nuclear binding energy of a helium atom is less than twice the nuclear binding energy of two hydrogen atoms. As you note, this goes as far as iron because this pattern is no longer true (ie the heavier fused element has more binding energy than the two lighter elements combined). I believe this particular conservation is "solved" or at least consistent with established theory, specifically QCD.
4. Fusion in stars is highly localized. Think of the volume of the Sun compared to the volume of every point in space to which the Sun was the closest star (ie approximately 2 light years in radius). It is astronomically small. To may layman's intuition it seems like this would be measurable in some way.
We know spacetime curves (based on the energy and mass present). We have detected the compression of space (ie gravity waves). We also know there is a minimum distance between objects (ie the Planck length) but (AFAIK) space isn't discrete as that might suggest.
There are huge gaps in our understanding of any of this.
I have no idea what im doing, i just like to take a idea and run with it.
But that idea could be tested, the more flat the space becomes, the more "positional" push-back by the bubbling space a space faring object would experience.