chore(examples): add vesseld-multi-vessel walkthrough

[lIang70]

Summary

PR-4 (final) of the vessel/vesseld v0.1 series. Documentation-only — no code, no go.mod changes, no CI changes.

Worked example demonstrating the YAML schema and HTTP control plane of vesseld v0.1.

What's in here

`examples/vesseld-multi-vessel/` — two independently-configured vessels sharing one OpenAI `LLMProfile` inside one daemon process:

File	Concept
`daemon.yaml`	Daemon-wide config: control socket, drain timeout, LLM rate-limit bucket.
`shared/openai.yaml`	One `LLMProfile` referenced by every vessel — the daemon shares the underlying client + limiter across them.
`vessels/triage/vessel.yaml`	A multi-agent vessel with dispatcher + worker + `Kanban` agent-as-tool delegation.
`vessels/support/vessel.yaml`	A simple single-agent vessel for plain Q&A.

Runnable via

```bash
go build -o ./vesseld ./cmd/vesseld
./vesseld validate --config examples/vesseld-multi-vessel -R
./vesseld start --config examples/vesseld-multi-vessel -R
```

The README walks through validate / plan / synchronous call / async submit + SSE log tail / drain.

Test plan

• YAMLs are byte-identical to `tests/e2e/vesseld/testdata/multi-vessel/` (PR test(vessel,vesseld): quality + e2e black-box suites #64) — same fixtures, same expectations.
• `vesseld validate --config examples/vesseld-multi-vessel -R` exits 0 against tip-of-main `cmd/vesseld`.
• CI green

Made with Cursor