> * Log queries where the optimizer row estimates were dramatically
>   different from the number of rows actually found (?)