Simple - the 'self-publishing' repository websites could work on the principle that unless more than 50 or 100 copies is sold per month it gets pulled. They could give new publications 6 months to achieve that level of sales. It won't make much difference for people who genuinely want to self publish via their own website but will prune the pointless dross that no one wants to read.
Its effectively what happens on a larger scale with real publishing - once sales drop the printing stops.
That would kill scientific publications as many are only read by a handful of people, but form the foundation of later work.
I'm sure you didn't mean that though. But setting levels wouldn't really benefit anyone, except perhaps the distributor.
Some books are very difficult to publish, even through self-publishing, and really need someone with a bit of muscle to throw their support behind. Some people have been thrown in prison for self-publishing their work. e.g. Fritz Springmeier (I've seen an old printing of his book selling on Amazon for hundreds of dollars - they're now down to less than $200 because it's back in print through Infowars).
The trouble, as I see it, is not that real publishers are dying in the new world because of self publishing but that the mega-corps are effectively ousting them - eventually Amazon, Apple, et al will become the new world publishers with even less help and support for authors as they establish a leech-like grip on authors.
Amazon and Apple are not really good for publishing overall. Apple is a well-known, belligerent censor, and Amazon is well-known for being ruthless. I have faith in that smaller publishers and self-publishing will continue for a long time to offer alternative perspectives. The philosophical perspective of many small publishers will prevent Amazon or Apple from swallowing them up, because neither Big A can stomach what a lot of people think.
But, I think you're right for a large sector of publishing. The Big A's will eat up most. Yay. Aren't we lucky...