feat(cli): structured output and agent-friendly error messages for fleet CLI

[khang859]

Overview

As Fleet adds more autonomous agent roles (Navigator, First Officer, etc.), the fleet CLI needs richer, more structured output across all commands — not just the Navigator-specific paths. This is a standalone improvement to make the CLI more reliable for smaller models (Haiku) and easier to build on top of.

Goals

• Consistent structured JSON output mode across all command groups
• Actionable error messages with hints and corrective suggestions (not just failure strings)
• Input validation with clear pre-flight errors before executing irreversible commands
• Next-step hints in successful command responses to guide agents toward correct follow-up actions

Scope

All existing command groups:

• fleet crew *
• fleet missions *
• fleet comms *
• fleet sectors *
• fleet cargo *

Out of Scope

Navigator-specific CLI output (structured output for FLEET_NAVIGATOR=1 env var, execution-scoped commands, protocol step guards) is handled as part of the Protocols & Navigator feature.

Notes

• Smaller models like Haiku won't recover gracefully from vague errors — the CLI needs to do more of the thinking
• Pattern: { "ok": true, "data": {...}, "hint": "suggested next command" } and { "ok": false, "error": "...", "hint": "corrective action" }
• Irreversible commands (crew recall, crew deploy) should validate all inputs upfront before executing