[llvm] [RISCV] Add TuneNoDefaultUnroll to generic CPUs (PR #135318)

[Pengcheng Wang via llvm-commits]

wangpc-pp wrote:

> I think we should revert this for now so it's easier to track other performance changes without the regression on main.
> 
> Btw, I think this is an issue with SLP, and it's also affecting [#195886 (comment)](https://github.com/llvm/llvm-project/pull/195886#issuecomment-5089708933) because that too causes SLP to run after unrolling.
> 
> The unrolling causes SLP to start grouping entries in a weird order which makes it think vectorization is unprofitable. I'll try and create a reproducer for this.

I will do that.

https://github.com/llvm/llvm-project/pull/135318