test(iroh): Make endpoint_relay_connect_loop not flaky

[matheus23]

Description

endpoint_relay_connect_loop was flaky before. This should fix this issue. It was last marked flaky (but for windows only) in #3354.

What seems to have happened:

• The test spawns a relay server
• The test spawns a server iroh Endpoint
• The server Endpoint does a net report with QAD probes
• For some reason the spawned relay server is very slow in responding to QAD requests (>3s)
• The QAD probes time out, the server Endpoint ends up without a home relay
• The test spawns a client iroh Endpoint
• The client Endpoint tries to connect for 30s
• The server Endpoint doesn't do another net report for 30s, though, so never ends up being reachable
• The client Endpoint times out.

To work around this, I'm starting the server endpoint and waiting for it to have a relay address.
IMO this is reasonable to do in tests.

I've also made the tests use Connection::close and Connection::closed properly and removed SendStream::stopped and RecvStream::read_to_end(0) calls.

Notes

There's some other drive-by changes. Sorry about that, but IMO they're kinda too small for their own PRs:

• net_report thought that Watchable::set would return Err when there's no more watchers listening, but that's incorrect: It returns Err from set when the value set is the same as the currently stored value.
• I've also made some small cosmetic changes to net_report
• I've removed a 1600 bytes allocation from the hot path of receiving relay items in ActiveRelayActor.

Change checklist

• Self-review.

[github-actions]

Documentation for this PR has been generated and is available at: https://n0-computer.github.io/iroh/pr/3402/docs/iroh/

Last updated: 2025-07-25T10:15:13Z

[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: 3c0d6e7