Add reasoning effort level to status line variables

Feature Request

Add a reasoning_effort (or similar) variable to the status line system so users can see what effort level is being applied per-message.

Motivation

When using effort: auto, the system dynamically selects an effort level per request. Currently there's no way to observe what level was chosen — the only visibility is /effort which shows the setting (e.g., "auto (currently medium)") but not the actual per-message value.

Power users who want to evaluate whether auto is making good decisions need to see the actual effort applied to each response. This is especially useful for:

• Calibrating auto behavior — understanding when auto chooses low vs. high effort
• Cost/latency debugging — correlating response quality with effort level
• Workflow tuning — deciding whether to override auto for certain task types

Proposed Solution

Add a status line variable like reasoning_effort.current (or effort.level) that reflects the effort level used for the most recent response. Example values: the numeric level (e.g., 85) or the bucketed label (low/medium/high).

Example status line config:

{
  "statusLine": "effort: {reasoning_effort.current} | ctx: {context_window.used_percentage}%"
}

Alternatives Considered

• Asking the model directly ("what effort are you at?") — works but wastes tokens and context
• /effort command — shows the setting, not the per-message actual
• Inferring from response length/depth — unreliable

Environment

• Claude Code CLI
• Using effort: auto setting