Expose LLVM instruction builder for `neg` and `fneg`

[JarnaChao09]

Purpose

Exposes the LLVM-C API to build neg and fneg instructions

Reason

While the neg instruction is just a "super instruction" to 0 - self, fneg as an instruction maintains NaN-ness inside of the floating point representation which may not be kept by performing 0 - self.

Notes

• If requested, an issue can be opened under the general goal of exposing more/updating the LLVM-C API to its modern counterparts (e.g. LLVMArrayType vs LLVMArrayType2)
• Bindings were tested as part of another project