How should I disclose a data extraction error when submitting a revision?

It is probably better to point it out. A note to the editors/reviewers is fine. You and everyone want the best science to be published. Full disclosure is best.

I doubt that this would lessen the chance of acceptance, but it might draw a bit more scrutiny overall. But that is a good thing in general.

This is similar to pointing out other changes in the paper, those that were called for and those that you think are relevant to improving the paper.

In your response to reviewers, you can add a section at the end titled "Other changes" or something like that. Explain this data extraction error and mention that it does not change the results to any noticeable extent.

I think that by putting it at the beginning, you're implicitly saying that it's a bigger issue than it really is. If it really does not invalidate the peer review process thus far, it is more a footnote than a main point.

I would also triple-check your other data extractions. While one slip-up is unfortunate, two might signal that you're not careful in your methodology. You could specify that you have verified all your results in your response as well, if you fear that reviewers would doubt your results.