Show AI credit consumption per automation run

### Feature summary

Surface AI credits (and token counts) consumed by each automation run in the GitHub Copilot app.

### What problem are you trying to solve?

Automations can be triggered frequently and run unattended, but the app does not show what an individual run consumed.

Run frequency is the main decision I have to make when setting up an automation, and cost per run is the input I need to make it. A run that costs 20 AI credits and a run that costs 400 AI credits imply completely different schedules. Today the only way to find out is to enable the automation, watch the total bill move, and reverse-engineer the per-run cost.

The practical result is that I under-use automations, because running them at a useful frequency feels like an unbounded financial risk.

### Proposed solution

- Show AI credits consumed for each run in the run detail view, alongside duration and status. Token counts per model would be a useful addition, but credits are the figure that maps to cost.
- Add a usage column to the automation's run history so runs can be compared at a glance.
- Show aggregate figures per automation: total and average credits per run over a selectable period (e.g. last 7/30 days), so the monthly cost of a given schedule can be projected.

### Workflow impact

Affects anyone running automations on a schedule or a frequent trigger, particularly individual users and small teams on usage-based billing where spend is directly visible.

With per-run cost visible, choosing a trigger frequency becomes an informed decision instead of a guess. It also makes it possible to spot an automation whose cost has drifted upward after a prompt or repository change.

### Installation context

_No response_

### Additional context

_No response_

No se nombran archivos, pruebas ni puntos de entrada. Empieza por localizar los detalles de ejecución de la automatización, el historial de ejecuciones y las vistas de uso agregado, y por rastrear los datos disponibles para cada ejecución. Se considera terminado cuando los créditos por ejecución son visibles, con recuentos de tokens opcionales y cifras totales y promedio para un período seleccionable.

Escrito por el modelo de indexación a partir del texto del issue.