added trivial malloc fast-path benchmark

While this is not good representation of real-world production malloc
behavior, it is representative of length (instruction-wise and well as
cycle-wise) of fast-path. So this is better than nothing.
3 files changed