Add bind-address-4 to upnp

[LaserEyess]

Make sure the bind-address-4 address is being passed to upnp. This does not affect ipv6 (which shouldn't need either UPnP or NAT-PMP).

Fixes #804, but does not solve the issue for NAT-PMP, which requires knowledge of the gateway. If there is a way to do that properly, and cross platform, I think that should be another PR.

Based on https://trac.transmissionbt.com/ticket/5990

[LaserEyess]

@ckerr @mikedld Could I get a review? Despite the lack of feedback from the person who made #804 this PR seems to correct the specific iffue of UPnP packets leaking on ipv4.

[LaserEyess]

@ckerr rebased this patch for the C++ change. I need to test that it's still working, but otherwise I think it's ready for review.

[ckerr]

It's failing to build in CI:

port-forwarding.cc:98:106: error: no member named 'public_ipv4' in 'tr_session' [clang-diagnostic-error]
tr_upnpPulse(s->upnp, private_peer_port, is_enabled, do_check, tr_address_to_string(&s->session->public_ipv4->addr));
~~~~~~~~~~ ^

[LaserEyess]

Well that's on me for not testing even that it compiles, I apologize.

As for actually testing the feature, as far as I can tell it seems to be working. When I set the bind addr to 10.1.1.1 (a wireguard interface) I get that it fails with Required key not available, which is expected since the wg interface can't route it, but I do not see packets going out my eth0 interface. When set explicitly to eth0, it works as expected. Additionally, while there was talk in #804 about a crash on an invalid IP address, that doesn't actually happen.I can set the address to "not.an.ip.address" and it still works.

[LaserEyess]

Also, I was unable to commit without adding TR_SKIP_HOOKS=1, I kept getting this error

checking code format
./libtransmission/quark.cc:415:19: error: code should be clang-formatted [-Wclang-format-violations]
    []() constexpr
                  ^
C/C++ code needs formatting

Yet I never touched that file, I assume it's unrelated.

[ckerr]

checking code format
./libtransmission/quark.cc:415:19: error: code should be clang-formatted [-Wclang-format-violations]
    []() constexpr
                  ^
C/C++ code needs formatting

Yep, I've seen that one too. CI is running clang-tidy 12, but clang-tidy 13 sees quark.cc:415 differently. Unrelated to this PR.

FTBFS has been resolved