I think then arXiv would have to deal with mantaining the tech stack and providing the presumably much higher server capacity to serve the more varied web pages that would result, so it seems like a tall order. arXiv already has an experimental integration with Papers with Code [0], which I guess provides similar results for the reader, though the authors have to figure out their own web hosting.
Second that. Something I put out recently had an (admittedly video heavy) webpage that had 1TB of traffic over the past month. Cloudflare handled it for free for me, but at ArXiv’s scale it’s bound to be a problem.
[0] https://info.arxiv.org/labs/showcase.html#arxiv-links-to-cod...