(interfaces.lib.inc) regression, parse "inet6 2001:470:xxx:xxx::2 --> 2001:470:xxx:xxx::1 prefixlen 128"

Nothing to gain from clustering too much.  The upside is that
we've got to clean up the settings for firewall, general and
misc considerably.

(firewall, nat) detect current configured address family, https://github.com/opnsense/core/issues/807

(firewall, nat) change default for none to empty, https://github.com/opnsense/core/commit/610013dcf37590350b7e41a32eaaccda08aa09ac#commitcomment-17585636

... the utmost potential to implode your setup.

Revert "(configd) send received output on script_output type when there is any, closes https://github.com/opnsense/core/issues/951"

This reverts commit b3049270.

(legacy) filter_log.inv move convert_real_interface_to_friendly_descr out of parse loop, https://github.com/opnsense/core/issues/672

(dnsmasq) add custom port option for domain overrides, closes https://github.com/opnsense/core/issues/799

(interfaces/dhcpv6 client) add debug option to advanced dhcpv6 client config, closes https://github.com/opnsense/core/issues/859

(firmware) move tooltip to button, apply tooltip for dynamic generated buttons. closes https://github.com/opnsense/core/issues/830

Requires fiddling with /etc/fstab, at least that is a file
that we never touch.  It's harder for installs that need
TRIM to be off all the time as the first boot does actually
enable it.  That means after the install is complete it needs
to be CTRL+C'ed and the fstab entry modified before reboot.

PR: https://forum.opnsense.org/index.php?topic=3044

NPT is a kind of abandoned feature that had a custom kernel patch.
I tried to trace the origin through OpenBSD, but their pf(4) since
switched away from separate binat rules in the last known form like
it is still in FreeBSD.

Furthermore, the original GUI commit looks odd in that it tries to
push the same traffic downwards that would match in the former line,
which acutally points upwards.  It's either that or completely zapping
the line.  For now, repair the rules reload by trying to retain the
spirit of what it tries to achieve and wait for further feedback.

This late bug report also suggests that virtually nobody uses NPT
today since we've had no upstream reports since at least 15.7 was
out.

PR: https://forum.opnsense.org/index.php?topic=3076.0

(configd) send received output on script_output type when there is any, closes https://github.com/opnsense/core/issues/951

The original script is from 2005 and most newer ath(4) drivers do not
even support it.  One can still set the sysctls individually anyway,
it's highly likely you know the hardware when you're trying to set
this up in the first place...

See also: https://wiki.freebsd.org/dev/ath(4)

Looks like this is working as intended now, shall be shipped with
16.1.16 after an extra round of testing.

Kindly pointed out by a generous tester.  Thank you.