fix(iroh): Use valid available IPv6 address, ignoring have_ipv6

[matheus23]

Description

Sometimes net_report says udp_v6: false, even though I can clearly send and receive to/from the other iroh endpoint I want to talk to.
Here are some logs from that situation:

2025-07-29T10:15:43.281088Z DEBUG ep{me=68b3f4f4cc}:actor: iroh::net_report: generated report in 728ms report=Report { udp_v4: true, udp_v6: false, mapping_varies_by_dest_ipv4: Some(false), mapping_varies_by_dest_ipv6: None, preferred_relay: Some(RelayUrl("[https://euc1-1.relay.n0.iroh.iroh.link./")](https://euc1-1.relay.n0.iroh.iroh.link./%22))), relay_latency: RelayLatencies { ipv4: {RelayUrl("[https://aps1-1.relay.n0.iroh.iroh.link./"](https://aps1-1.relay.n0.iroh.iroh.link./%22)): 182.188896ms, RelayUrl("[https://euc1-1.relay.n0.iroh.iroh.link./"](https://euc1-1.relay.n0.iroh.iroh.link./%22)): 21.317467ms, RelayUrl("[https://use1-1.relay.n0.iroh.iroh.link./"](https://use1-1.relay.n0.iroh.iroh.link./%22)): 117.29417ms}, ipv6: {}, https: {} }, global_v4: Some(176.199.xxx.xxx:26608), global_v6: None, captive_portal: None }
2025-07-29T10:15:55.933530Z DEBUG ep{me=68b3f4f4cc}:disco_in{node=e10b5cdea1 src=Ip([2a01:599:b2f:a1ad:495a:8b14:xxxx:xxxx]:49154)}:handle_disco{dm=Pong(Pong { tx_id: TransactionId(0xBC381DA98EFD2B20CAE0A96E), ping_observed_addr: Udp([2a02:8071:d85:19c0:9676:965:xxxx:xxxx]:38391) })}:handle_pong{m=Pong { tx_id: TransactionId(0xBC381DA98EFD2B20CAE0A96E), ping_observed_addr: Udp([2a02:8071:d85:19c0:9676:965:xxxx:xxxx]:38391) } src=Udp([2a01:599:b2f:a1ad:495a:8b14:xxxx:xxxx]:49154)}: iroh::_events::holepunched: remote_node=e10b5cdea1 path=[2a01:599:b2f:a1ad:495a:8b14:xxxx:xxxx]:49154 direction="outgoing"

Now, the udp_v6: false value originally comes from netwatch's InterfaceState, as it sees my (what is now and probably was at the time fe80-prefixed) link-local address and characterizes it as "not reachable from the public internet".

But this PR instead ignores that value iff we have any UdpSendAddr::Valid, since that would mean it's an address that we've successfully sent and received from with the particular peer we care about.

Change checklist

• Self-review.

[github-actions]

Documentation for this PR has been generated and is available at: https://n0-computer.github.io/iroh/pr/3419/docs/iroh/

Last updated: 2025-08-01T09:44:29Z

[github-actions]

Netsim report & logs for this PR have been generated and is available at: LOGS
This report will remain available for 3 days.

Last updated for commit: 71d77c2

[matheus23]

Unfortunately further testing didn't have the intended result: I wasn't able to reproduce the case I had above where I was magically able to send/receive IPv6 traffic to someone else's home address, but not to the relays (which I agree is very unusual).

In any case, I still believe this is a valid change. If we end up for some reason having a valid IPv6 addr, we should probably use it.
If it's the case that it's only valid because we had IPv6 in the past, but not in the future, then it'll become invalid after 6.5 seconds and we'll use a valid/outdated/unconfirmed IPv4 address instead then.