`pandas` does not do this. Instead of using `pd.merge` you could have used a DataFrame merge and in fact just like `polars` constructed a single expression to do the same thing.