benchmark inconsistency in notebook on profiling and optimizing

[declann]

A little inconsistency in the notebook on profiling and optimizing the savings model:

Table and text are inconsistent. There is still a significant performance improvement when I run now: but it's more like ~16% down to ~8% (with a fair amount of variance)

[fumitoh]

Thanks! Would you care to PR? (Updating the text and rerunning the whole thing.)

[declann]

Can do soon! (when I get around to do some environment setup)