I added a QL-for-QL query that flags suspicious QLDoc for a predicate when:

• At least one parameter is documented in the QLDoc.
• At least one paramter is not documented in the QLDoc.
• The QLDoc has a codeblock (enclosed in backticks) mentioning something that looks like an identifier and is not a parameter.

This turns out to be good at identifying lots of QLDocs where the parameters mentioned in the qldoc do not match up with the actual parameter names.

I fixed most of the issues identified in the javascript/ folder. The remaining issues in javascript/ seem benign.
There are still lots of issues for the other languages. I have not tried to fix those.

Could this be tuned to handle the case where all parameters have been renamed? That case can not be caught currently due to the first of your requirements. (I'm particularly interested in the /1 case)

Could this be tuned to handle the case where all parameters have been renamed? That case can not be caught currently due to the first of your requirements. (I'm particularly interested in the /1 case)

I thought it would be bad, but it worked great 👍 (See the extra JS fixes).

I also found a few more tweaks to reduce the FP rate.
There are also more benign cases, but I think the precision is OK.

Edit: And I did some drive-by patching of explicit-this and redundant-cast, to further reduce the number of results in javascript/.