Expose context category breakdown in statusLine JSON input

Feature Request

The statusLine.command script receives a JSON blob with token usage data, but it only includes aggregate numbers:

• context_window.current_usage.input_tokens
• context_window.current_usage.cache_read_input_tokens
• context_window.current_usage.cache_creation_input_tokens

The built-in /context command already computes a per-category breakdown:

• System prompt
• System tools
• MCP tools
• Custom agents
• Memory files
• Skills
• Messages
• Autocompact buffer

Request: Include this per-category token breakdown in the JSON blob passed to statusLine.command scripts, e.g. under context_window.categories:

{
  "context_window": {
    "categories": {
      "system_prompt": 6600,
      "system_tools": 9100,
      "mcp_tools": 341,
      "custom_agents": 3000,
      "memory_files": 3700,
      "skills": 3500,
      "messages": 51300,
      "autocompact_buffer": 33000
    }
  }
}

Use case: Custom status lines that show context composition at a glance without needing to run /context interactively. This enables users to monitor which categories are consuming context and optimize accordingly (e.g., reducing MCP tools or skills when approaching limits).