In the SIR model, a factor 1-1/R0 of the total population end up being infected over time. That means with a reduction of 0.1, you will prevent about 0.1/R0^2 from ever being infected, so something like 2%. Of course there are other effects like less load on the hospitals.

Yes, but if you look closely at italy, that's what matters: spread the amount of people requiring intensive care until it is below the threshold your region is able to provide.

That's right. Even if it is impossible to prevent from spreading, which I believe we will no longer be able to do, it is still much better to be sick at some other time than when everybody else is sick, too.

It also gives more time to develop protocol to treat it.

Exactly, as https://www.flattenthecurve.com/ puts it, it's all about reducing the load on the healthcare infrastructure:

"Far and away, the most important thing to do is flatten the curve of the epidemic so that our health systems can cope and to give time for the scientists to research vaccines and treatments."