fix: drain outbound connections during shutdown

[GrapeBaBa]

Description

Prevent shutdown from finishing while outbound dial jobs or TCP sockets still own referenced native handles.

• mark the connection manager stopped before service teardown
• abort and await active dial jobs
• track outbound TCP sockets and wait for their close events
• avoid resetAndDestroy after graceful socket shutdown has started, since libuv does not allow mixing uv_shutdown with uv_tcp_close_reset

This was observed while investigating a Lodestar shutdown hang: https://gist.github.com/nflaig/5f41cfc50f38baf5046a034162943dc3

Tested version

This draft intentionally points to commit c20c703, the exact js-libp2p source used to build Lodestar image sha256:2c2bcfbe1a9adf72cb9f73dfc605fb3a149e283d8aeae6b0325bbefd5e1b13f9.

The affected compiled libp2p and TCP files in the running image were compared byte-for-byte with the artifacts built from this commit. All hashes match.

The image has completed 44 consecutive effective mainnet shutdown cycles without a hang so far. The 50-cycle soak is still running.

This branch is intentionally not merged with newer main commits yet, so the PR head remains identical to the tested version.

Verification

• Added real TCP regression coverage for outbound socket close and the reset/graceful-close race.
• At this exact commit, the affected package lint, doc check, build, and 39 Node.js tests passed.
• No tests were rerun while opening this draft PR.

Change checklist

• I have performed a self-review of my own code
• I have made corresponding changes to the documentation if necessary (this includes comments as well)
• I have added tests that prove my fix is effective or that my feature works

[GrapeBaBa]

Superseded by #3597. A controlled 50-cycle mainnet shutdown soak showed that only the minimal socket reset guard was required: unmodified Lodestar, 50 clean exits, 0 hangs.