feat: Add --json flag for LLM-friendly CLI help output

[aronchick]

Summary

• Adds --json flag to CLI help for machine-readable output
• multiclaude --json outputs full command tree as JSON
• multiclaude --help --json same as above
• multiclaude <cmd> --json outputs that command's schema

This enables LLMs and automation tools to programmatically discover available commands, their descriptions, usage patterns, and subcommands.

Implementation

• Added CommandSchema struct for clean JSON serialization
• Added toSchema() method on Command that recursively converts the command tree
• Internal commands (prefixed with _) are filtered from output
• Updated help text to mention the new --json flag

Example Output

$ multiclaude agent --json
{
  "name": "agent",
  "description": "Agent communication commands",
  "subcommands": {
    "attach": {
      "name": "attach",
      "description": "Attach to an agent's tmux window",
      "usage": "multiclaude agent attach <agent-name> [--read-only]"
    },
    ...
  }
}

Test plan

• Build succeeds: go build ./cmd/multiclaude
• All tests pass: go test ./...
• New tests added for JSON output functionality
• Manual verification of --json flag at root and subcommand levels
• Regular help output still works as expected

🤖 Generated with Claude Code

[aronchick]

✅ Merge Queue: READY TO MERGE

• CI: All 6 checks passing
• Merge status: MERGEABLE/CLEAN
• Reviews: None blocking

This PR is ready for maintainer merge.