gh-131798: Use sym_new_type instead of sym_new_not_null for _BUILD_STRING, _BUILD_SET

[Zheaoli]

This PR is under developing, not ready for review

• Issue: Better uop coverage in the JIT optimizer #131798

[Zheaoli]

@brandtbucher Hi Brandt, I might need some help here.

I have already optimization path for _BUILD_STRING(I think we just need setup the const after we check all the opargs string on stack.

But I got a little bit confused for _BUILD_STRING. If I'm correct, _BUILD_SLICE is equal with BUILD_SLISE, I'm not sure it's necessary add an optimization path for _BUILD_SLICE. The result is const or not is not related with the argument?(not sure about this)

[brandtbucher]

Hi @Zheaoli! Let's start by just setting the output types for these. I think maybe the constant output stuff may not be needed after all, due to GH-132733.