Things I wanted to tweak
- log the benchmark results to a logger instead of system.out so the output doesn't get lost
- split the bench variable to specific tests, so you can choose which one you want to run
- cleanup the imports
- cleanup of test directories simply doesn't happen
- revisit the default values (ex. benchLargeMap has the requirement of 1G but goes all the way up to 1000 iterations, so 1g is no where near enough heap).