This PR improves the inlining for allocations considerably by ensuring that the bump allocator gets inlined into the code generator. We do this by means of moving kore_arena_alloc to the header file and making it an inline definition, as well as modifying the build process so that the bitcode file generated by alloc.cpp gets linked using LLVMs bitcode linker into the code generator module prior to optimization.
This PR improves the inlining for allocations considerably by ensuring that the bump allocator gets inlined into the code generator. We do this by means of moving
kore_arena_allocto the header file and making it an inline definition, as well as modifying the build process so that the bitcode file generated by alloc.cpp gets linked using LLVMs bitcode linker into the code generator module prior to optimization.