fix: gracefully handle AddrInUse error instead of panicking on startup

[sanity]

Problem

When starting freenet network while port 7509 is already in use (e.g., another freenet process is running), the node panics with an unhelpful stack trace:

thread 'tokio-runtime-worker' panicked at .../src/server/mod.rs:72:68:
called `Result::unwrap()` on an `Err` value: Os { code: 98, kind: AddrInUse, message: "Address already in use" }

This panic cascades through the system, making it difficult to diagnose the actual problem.

Solution

Changed serve() to bind the TCP listener synchronously before spawning the async server task. This allows proper error propagation:

1. Bind listener first (before spawning task)
2. Return std::io::Result<()> instead of ()
3. For AddrInUse errors, provide a clear message:

   "Port X is already in use. Another freenet process may be running. Use 'pkill freenet' to stop it, or specify a different port with --gateway-port."

The key insight is that the original code spawned a task and then tried to bind - so the error happened asynchronously after serve() returned, making it impossible to propagate to the caller. By binding before spawning, errors surface immediately and can be handled properly.

Changes

• serve(): Now async, binds listener before spawning, returns Result
• serve_gateway(), serve_gateway_for_test(), serve_gateway_in(): Return std::io::Result
• All callers updated to propagate the error with ?

Testing

• cargo check --all --tests passes
• cargo clippy -p freenet --all-targets passes

Closes #2237

[AI-assisted - Claude]

[iduartgomez]

✅