Well the reason to auto-close is that this is not an entirely unlikely resolution. When I inherited a project with a bunch of issues and started going through the backlog, around 50% of tickets were duplicates, already solved, unreproducible, etc etc

When you've only got limited time, having less of those issues to analyze and then close anyway is a very valid reason. It leaves more time for fixing real issues. Of course it comes at the cost of ignoring perfectly valid issues as well, that's why this is obviously never an optimal policy to implement, and should only be done in desperate situations.