Display underlying LLM model used by custom agents in UI

### Describe the feature or problem you'd like to solve

Currently, when utilizing a multi-agent architecture with custom agents (e.g., `orchestrator`, `planner`, `coder`, `security-reviewer`), there is a lack of transparency regarding which specific LLM model each agent is using during execution. When observing the terminal/UI output, I can see the agent's name and the actions it is taking (e.g., `One-piece:coder Verify call CTA`), but there is no indication of the underlying model powering that specific agent.

Furthermore, I have checked the following areas to verify my configuration, but the model information is completely missing:
1. **Session Logs:** The logs do not contain any metadata indicating which model was used for specific agent prompts/responses.
2. **The `/tasks` command:** Running this command yields no relevant information regarding agent-to-model mapping.

### Proposed solution

I would like to see the assigned model surfaced in a few key areas for debugging, cost-tracking, and transparency:

1. **In the Execution UI/Terminal:** Append or display the model name next to the agent name.
*(Example: `One-piece:coder [claude-3.5-sonnet] Verify call CTA`)*
2. **In the Session Logs:** Include the model name in the payload/metadata of the logged interactions.
3. **In the `/tasks` command (or a new command like `/agents`):** Provide a summary table showing active agents and their currently configured models.

### Example prompts or workflows

_No response_

### Additional context

Here is a screenshot of the current UI. As you can see, `One-piece:coder`, `security-reviewer`, and `code-reviewer` are working, but it is impossible to tell which models they are utilizing for these tasks.

La issue identifica l’interfaccia utente di esecuzione o l’output del terminale, i log della sessione e il comando /tasks come punti di ingresso interessati, ma non indica file o test. Inizia tracciando come viene visualizzata l’attività di custom-agent e come vengono prodotti i metadati di interazione e i dati di /tasks; il lavoro è completato quando il modello dell’agente selezionato è visibile in modo coerente nell’interfaccia richiesta, nei log e nel riepilogo dell’agente.

Scritto dal modello di indicizzazione a partire dal testo della issue.