"powertop" reports fail2ban as having 30 wakeups per second
(with not that many files to check!).

This might not matter much for a big beefy busy server;
on an otherwise mostly idle system it's annoying.

(Reported in "top"s output as using 1-3% cpu all the time
on an rpi3, arm64).

I can confirm this on my bullseye test system:
- powertop: Multiple fail2ban threads with up to 12 events/s
- top: 0.3-1.0% CPU usage