[llvm] [IR] Add intrinsics to represent complex multiply and divide operations (PR #68742)

[Eli Friedman via llvm-commits]

efriedma-quic wrote:

I dug a bit into academic sources around the rounding... apparently it's a well-studied problem. If anyone else is interested in going down the rabbit hole, see https://inria.hal.science/inria-00120352v2/document ("Error Bounds on Complex Floating-Point Multiplication") and https://ens-lyon.hal.science/ensl-00649347v4/file/JeannerodLouvetMuller13.pdf ("Further analysis of Kahan’s algorithm for the accurate computation of 2 x 2 determinants").

https://github.com/llvm/llvm-project/pull/68742