This code here isn't representative of any Haskell end-users would write, it has a whole other syntax (LiquidHaksell) layered on top of it that augments the typ

This code here isn't representative of any Haskell end-users would write, it has a whole other syntax (LiquidHaksell) layered on top of it that augments the type system and it imports the internals of GHC.* libraries to work with the base machine types. The strength of GHC's implementation is that a lot of the compiler internals are directly accessible in the frontend source language with certain language extensions enabled (GHC.Exts, GHC.Prim, ``-XMagicHash``). That said the code here is more about the implementation of the vector library internals at a very low level that deals with machine types. If you're a Python user this is effectively equivalent to reading the NumPy internals which I can tell you first hand are an ungodly mess of preprocessor macros and code generation gone horribly wrong and far worse than this.