When nftables is restarted/reloaded, fail2ban must be restarted

[nis65]

the nftables restart/reload cleares the stuff created by fail2ban (e.g. table inet f2b-table).

This creates ugly interdependencies, maybe I should analyze this together with #30

[nis65]

• Idea 1: combine nftables and fail2ban into a single role so that only there nftables and dnsmasq are restarted -> does not scale: there are other roles adding files to nftables and need a restart: dnsmasq, openvpn and unifi
• Idea 2: combine nftables and fail2ban via systemd (when nftables is restarted, dnsmasq is too). --> Seems not to be obvious, but this link is 6 years old...

[nis65]

Found a solution that works

• make fail2ban unit PartOf nftables unit (this solves the systemd restart nftables issue
• make fail2ban unit restart (and not reload) even when systemd reload is called (a reload of the fail2ban unit does not restore the fail2ban nft datastructures after a nft reload. A restart does).
• make nftables unit propagate the reload to fail2ban

The behaviour is now

• If you restart/reload nftables, fail2ban is restarted
• I you restart/reload fail2ban, fail2ban is restarted

Not the most beautiful solution, but for my use case sufficient.