[Idea] Estimate of time to create tokens that correspond to scientific dataset like arxiv

[nickmvincent]

Include inputs such as researcher wages and avg # of paper produced per year, avg # of tokens in a paper, etc. and a corresponding scenario.