For the leak scenario, I envision taking two heapsnapshot files and comparing them, because that's the most convenient way to figure out what actually leaked
Fo the absolute memory usage scenario, I envision just adding the Chrome/Puppeteer flag that allows precise memory stats and then use window.performance.memory.
adding the ability to automatically run 1 iteration of each benchmark and measure the memory before/after (using special Chrome flag to expose precise memory) is reasonable
users can create their own leak tests by writing a test where it allocates a bunch of stuff and then de-allocates, so it would expect the diff to be ~0
putting chrome in precise memory mode should not affect CPU perf
It would be convenient if Best could measure not only CPU time, but also memory usage. I can envision two types of tests here: