Add raw token count fields to statusline input JSON

Feature Request

Description

Add raw token usage fields (e.g. usage.input_tokens, usage.output_tokens, usage.total_tokens) to the JSON that Claude Code pipes to the custom statusline script on stdin.

Motivation

Currently the statusline input JSON only provides cost.total_cost_usd, which shows dollar spend but not raw token counts. Users who want to monitor token consumption directly — for budgeting, tracking context usage trends, or building custom statusline indicators — have no way to access this data.

Proposed fields

{
  "usage": {
    "input_tokens": 12345,
    "output_tokens": 6789,
    "total_tokens": 19134,
    "cache_read_tokens": 8000,
    "cache_write_tokens": 2000
  }
}

Use case

Display raw token counts in a custom statusline, e.g.:

◈ Opus 4.7 | [■■■■■■■□□□□□□□□□□□□□] 35% | 19.1k tokens | ↯ high | ⎇ main

This would allow users to track their actual token consumption alongside (or instead of) the dollar-cost display.

Current workaround

None — the data is not exposed to the statusline script at all.