sync-fs mostly measures time calling the overriden Math.random function

[anba]

#19 updated "async-fs" to not be a benchmark for the overridden Math.random function. Can the same change to applied to "sync-fs"?

Replacing view[i] = (Math.random() * 255) >>> 0; with view[i] = i in randomFileContents shows that:

• V8 spends about ~75% of its time evaluating Math.random().
• SpiderMonkey spends about ~85% of its time evaluating Math.random().