mm/damon/paddr: drop last same folio access check reuse optimization

It can race when multiple kdamonds are being used. The problem from the race is doubtful, but the gain from the optimization is also doubtful. Simply drop the optimization in favor of code simplicity