[DOC] incorrect OpenSearch Score for innerproduct in knn scriptscoring

[tstadel]

What do you want to do?

• Request a change to existing documentation
• Add new documentation
• Report a technical problem with the documentation
• Other

Tell us about your request.
The formula for the OpenSearch Score of the innerproduct in the table at the bottom of https://opensearch.org/docs/latest/search-plugins/knn/knn-score-script/ is incorrect. According to this formula, an increasing innerproduct metric would lead to a lower OpenSearch Score, which is clearly the opposite of what is intended by the dotproduct similarity metric.

This is only a problem with the docs. In reality, OpenSearch correctly calculates the innerproduct OpenSearch Score for script scoring according to the formula in the table at the bottom of https://opensearch.org/docs/latest/search-plugins/knn/approximate-knn/ (adapted for the additional minus in the distance function).
Most likely this is a follow-up error regarding the once wrong innerproduct distance function reported in #397 (which once was the same as l1). It seems like the distance function has been corrected but the OpenSearch Score is still the one for l1.

What other resources are available?
related to #397