[llvm] [AArch64] Don't try to vectorize fixed point to fp narrowing conversion (PR #130665)

[David Green via llvm-commits]

https://github.com/davemgreen commented:

I've put together a patch in #132366 to update the costs. I will look into updating the GISel codegen too, if you are not looking into it. I am hoping it is just an update to the legalization descriptors for the operations.

https://github.com/llvm/llvm-project/pull/130665