Port RayTracer benchmark to SoA algorithm using Vector128/256<T>

[fiigii]

The current SIMD benchmark RayTracer is using Vector3 to vectorize the basic 3D vector operations. However, this program is an AoS algorithm that cannot leverage wider SIMD operations. So, we should port RayTracer to a SoA algorithm using AVX intrinsics with Vector256<T>, then extend it to SSE* intrinsic with Vector128<T>.

cc @CarolEidt @tannergooding @AndyAyersMS @jkotas

[fiigii]

I will work on this issue, please assign to me.

[AndyAyersMS]

Please make a new benchmark so we have both versions.

[fiigii]

Yes, this benchmark is only for Intel hardware intrinsic, which will be put in https://github.com/dotnet/coreclr/tree/master/tests/src/JIT/HardwareIntrinsics/X86

[AndyAyersMS]

Perhaps you should add it over where we have all the other benchmarks:

https://github.com/dotnet/coreclr/tree/master/tests/src/JIT/Performance/CodeQuality

[fiigii]

Thanks for the suggestion. Will do.