feat(cli): add token usage block to --json run report

[RONAK-AI647]

Description

While finalizing the NanoBench architecture, I identified a critical missing dependency: downstream evaluation harnesses have no way to programmatically track token consumption or diagnose API quota limits.

Following up on the discussion in Nano-Collective/docs#60, I investigated issues #756 and #796. Both issues highlight a clear community need for token telemetry. After reviewing them, I found neither of those addresses the --json run report. NanoBench's orchestrator runs Nanocoder in --plain --json mode and reads structured stdout — that's a completely different code path from the interactive CLI footer or the VS Code ACP layer.

So #796, as written, would not actually put a usage block into the payload NanoBench needs. It's adjacent (same underlying ResponseUsage concept, same provider-reported token data), but it doesn't close the gap mentioned in NanoBench: Nano-Collective/docs#60 (comment)

What this PR does

This PR extends the headless stdout payload to expose total token consumption metrics when Nanocoder is run with the --json flag.

Key Changes:

• Multi-Turn Accumulation: Tracks and sums input, output, and total tokens across the entire conversation loop in runPlainConversation.
• Schema Extension: Injects a structured usage block into the final --json output.
• Graceful Degradation: Safely omits the usage property for local/unsupported models that don't report token metrics, preventing parser breaks.
• Test Coverage: Adds unit tests in conversation.spec.ts and shell.spec.ts to validate state tracking and payload generation.

It captures the usage data from the underlying AI SDK provider and appends a usage block to the final JSON report. The shape looks like this:

{
  "kind": "run_completed",
  "exitCode": 0,
  "toolCalls": [...],
  "filesChanged": [...],
  "temperature": 0,
  "usage": {
    "inputTokens": 4520,
    "outputTokens": 850,
    "totalTokens": 5370
  }
}

Type of Change

• New feature

[github-actions]

No changeset found

This PR does not add a changeset, so it will not appear in the changelog or trigger a release.

If the change is user-facing, add one:

pnpm changeset

Pick a bump (patch / minor / major) and write the changelog entry in our usual voice ("Added X... Thanks to @you. Closes #123."), then commit the generated .changeset/*.md file.

If this PR is docs-only or a chore that needs no release note, you can ignore this - or run pnpm changeset --empty to record that intentionally.

[RONAK-AI647]

@will-lamerton , review it please !