Digital archival of PDFs weighing a few hundred KBs to a few MBs is definitely a solved problem. And there are already arXiv overlay journals out there, and platforms supporting them. Tim Gowers' (Fields medalist) blog posts on this topic are quite informative:

https://gowers.wordpress.com/2015/09/10/discrete-analysis-an...

https://gowers.wordpress.com/2019/10/30/advances-in-combinat...

Highlights: $10 per submission, plus some fixed costs, including archival with CLOCKSS. No Elsevier extortion ring needed.

Impact factors are of course kind of a chicken and egg problem. Need to have enough high profile journals move off Big Publishing, or have enough high profile ones started.

> When I was a grad student at MIT, it was easy to read papers; if your IP was from MIT, every paper was 1 click away.

When I was a grad student at <institution of similar caliber>, or an undergrad at <another institution of similar caliber>, accessing papers was rather painful off campus. One either has to use EZproxy, which might decide to block you if it doesn't like your IP range (say in a foreign country), or use some godawful proprietary VPN client that I would stay the hell away from unless necessary.