As a side-note, IPv6 gateway switching is really discouraged,
especially with auto-address configuration techniques...

Discussed with: @adschellevis

filter, add "reply-to" to plugin model, disabled by default for bootstrapped rules. needs gateway/gatewayv6 registered in filter.inc before it can actually function.

filter plugin, work in progress. add systemDefaults to plugin and add disablereplyto tag in there, so we can set this tag globally without having to link the config in the filter parsing logic.
This shouldn't hurt our current deployment, it's needed to add "reply-to" parsing in the rule class.

missed the pfctl counters in www/widgets/api/plugins/interfaces.inc, for https://github.com/opnsense/core/issues/1662

regression, missing ipv4 address in get_interfaces_info(), for https://github.com/opnsense/core/issues/1662

get_interfaces_info() only overwrite status code when reported status in 'active','running' for https://github.com/opnsense/core/issues/1662

interfaces, refactor and rename get_interface_info() to remove most of the performance bottlenecks. for https://github.com/opnsense/core/issues/1662

interfaces.lib.inc, extend ifconfig parsing in legacy_interfaces_details(), for https://github.com/opnsense/core/issues/1662

step one, cleaning up get_interface_info(), move pfctl calls to configd. for https://github.com/opnsense/core/issues/1662

Replace get_configured_ip_addresses() and get_configured_ipv6_addresses() with a more pragmatic approach, both are used to detect if an address is already configured, which gets quite sloppy when having a large number of interface
s configured.

This change probes all configured interfaces at once, and returns the full list, stripping quite some magic in the process.

Because the indexing is changed, we need to change the consumers as well here.

rc.recover doesn't work like that then, need to move to /bin/sh.

PR: https://forum.opnsense.org/index.php?topic=5213.0

If a field isn't required and the user sets it with a faulty value, it would pass because it didn't know it wasn't set before user input.
There where two issues, first one, not knowing it was in init state (solved with internalFieldLoaded), next all field types should have used actionPostLoadingEvent to customize loading behaviour.

There are UX issues with this anyway and the missing license
worries me too.