This comment has been minimized.
This comment has been minimized.Show comment Hide comment
Sorry for the late answer, I never noticed your comment.
As explained in my various blog posts, this table is using the "name" type, leading to really bad estimations. You have to filter out indexes that we can not have a good estimation using the "is_na" field.
I should add links to my blog posts as documentation or put these queries in a repository with a proper README at some point...