[clang] [Clang] Add float type support to __builtin_reduce_add and __builtin_reduce_multipy (PR #120367)

[Farzon Lotfi via cfe-commits]

farzonl wrote:

> I'd definitely suggest you reduce the scope - maybe always expand the reduction serially in the frontend and not create the llvm intrinsic at all? 

Between @Il-Capitano  and @RKSimon comments I think it makes the most sense to do a sequential reduction in the frontend. Thats actually easier because we won't have to add SPIRV backend support for vec reduce. 


https://github.com/llvm/llvm-project/pull/120367