Initially non-flat space-time makes dark matter obsolete$dots$

I think there a few problems with this idea.

First off, as Stan Liou indicates, the early universe expanded extremely rapidly, potentially through hyper-inflation during the inflationary epoch. While we're still looking for observational evidence that this period occurred (e.g., the failed BICEP2 results a few years ago), the existence of such an epoch early in the universe neatly explains away several problems with the standard Big Bang theory (see the horizon problem, monopole problem, and the flatness problem). Assuming such a period existed, then any region of space-time that was "crumpled" would have been quickly and cleanly smoothed out during the hyper-inflation. Such crumpled-ness would not have been around long enough to affect cluster growth and structure. What's more, we very clearly measure that the Universe is flat over large, cosmic scales so it seems unlikely that there are constant and pervasive wrinkles as you suggest there could be.

A second issue I potentially see with your proposal is that it would imply a top-down approach to galaxy formation. That is, first you get huge clusters of mass (conglomerated by your "crumpled" space-time), then you get fragmentation and individual galaxies can form, then galaxies fragment into individual stars, etc. In other words, big structures form first, then smaller structures derive from the bigger structures. The opposite formation model would be "bottom-up" formation where first small things clump, then those clump into larger bodies and soon you have galaxies clumping into clusters. I don't think there is definitive evidence that galaxy formation is either top-down or bottom-up, but simulations seem to me to suggest bottom-up is more likely (of course those simulations use dark matter). I'll leave you to think about that one.

The last point of contention I have is that dark matter plays more of a role than just providing extra gravity. It contributes to the total mass-energy of the universe. Current theoretical models perform extremely well when we include dark matter. I believe a strong piece of evidence for this fact is our ability to theoretically match the CMB power spectrum using a Lambda-CDM model. I don't see that crumpled space-time could reproduce this result nearly as well.

I think also you'd have a very very hard time describing a mechanism that could allow "crumpling" to exist in the space-time of our universe which exactly matches the observed gravitational effects in galaxies. We can easily see that they appear to have a halo of matter which follows something akin to an NFW halo or else an Einasto profile. If you can find away that crumpled space-time would naturally form to match these profiles and somehow not be smoothed out by hyper-inflation, I'm all ears.