[Network] Add simple queuing metrics for message sends.

[JoshLind]

Description

This PR adds simple queuing metrics for message sends to the network stack. Specifically, we now track the time between when a message is sent by an application (after serialization), until when it is finally written to the network wire.

The next set of PRs will include: (i) receiver side metrics; and (ii) support for messages with embedded timestamps for tracking transport times.

Testing Plan

Existing test infrastructure and manual verification, e.g., I hacked the code to force messages to be streamed. You can see the new metrics in the Network Queue Latencies section of the dashboard.

[trunk-io]

⏱️ 3h 57m total CI duration on this PR

Slowest 15 Jobs	Cumulative Duration	Recent Runs
test-target-determinator	1h	🟩 ⬜ 🟩 🟩 🟩 (+10 more)
rust-cargo-deny	27m	🟩 ⬜ 🟩 🟩 🟩 (+11 more)
check-dynamic-deps	25m	🟩 🟩 🟩 🟩 🟩 (+11 more)
rust-move-tests	10m	🟩
rust-move-tests	10m	🟩
rust-move-tests	10m	🟩
rust-move-tests	10m	🟩
rust-move-tests	9m	🟩
rust-move-tests	9m	🟩
semgrep/ci	9m	🟩 🟩 🟩 🟩 🟩 (+11 more)
rust-move-tests	8m	🟥
general-lints	7m	🟩 🟩 🟩 🟩 🟩 (+11 more)
rust-move-tests	7m	🟥
rust-move-tests	7m	🟥
rust-move-tests	7m	🟥

_{settings ⋅ feedback ⋅ docs ⋅ learn more about trunk.io}

[github-actions]

✅ Forge suite realistic_env_max_load success on 8a39b440229c6446a147f0fc5c2943ec981a723a

two traffics test: inner traffic : committed: 14506.93 txn/s, latency: 2743.94 ms, (p50: 2700 ms, p70: 2700, p90: 3000 ms, p99: 3600 ms), latency samples: 5516000
two traffics test : committed: 100.02 txn/s, latency: 2342.01 ms, (p50: 2100 ms, p70: 2100, p90: 2300 ms, p99: 11200 ms), latency samples: 1700
Latency breakdown for phase 0: ["MempoolToBlockCreation: max: 1.262, avg: 0.769", "ConsensusProposalToOrdered: max: 0.329, avg: 0.292", "ConsensusOrderedToCommit: max: 0.355, avg: 0.343", "ConsensusProposalToCommit: max: 0.644, avg: 0.636"]
Max non-epoch-change gap was: 0 rounds at version 0 (avg 0.00) [limit 4], 1.16s no progress at version 2115341 (avg 0.20s) [limit 15].
Max epoch-change gap was: 0 rounds at version 0 (avg 0.00) [limit 4], 8.55s no progress at version 2115339 (avg 7.17s) [limit 15].
Test Ok

• Grafana dashboard
• Humio Logs
• Axiom Logs
• Validator CPU Profile
• Fullnode CPU Profile
• Test runner output
• Test run is land-blocking