Added llvm based process triple partitioning

[kc611]

As titiled,

This PR aims to add a llvm based binding that uses the internal Triple object to partition the process triple string into it's respective parts.

Example:

from llvmlite import binding

triple_string = binding.get_default_triple()
triple_string = "armv6hl-none-linux-gnueabi"

triple_parts = binding.get_triple_parts(triple_string)

print("Triple String:", triple_string)
print("Triple Parts:", triple_parts)
# Triple String: armv6hl-none-linux-gnueabi
# Triple Parts: ('arm', 'v6hl', 'unknown', 'linux', 'gnueabi', 'ELF')

[gmarkall]

Following the addition of the above change, I wonder if it's worth also adding tests that check with the XCOFF, Wasm, MachO, and GOFF cases too (there seem to be no tests in the LLVM test suite that result in an unknown object format).

[gmarkall]

Source for the above object format enum is: https://github.com/llvm/llvm-project/blob/f28c006a5895fc0e329fe15fead81e37457cb1d1/llvm/include/llvm/ADT/Triple.h#L237-L246