As the author, I totally agree. Especially for the deduplication benchmark there are so many variables that affect it that is definitely is worth restating.
That being said, I am quite confident the results are accurate and don't mind other people running their own tests.