feat: per-response token + speed display after model reply (closes #17)

[VforVitorio]

Status

Implemented in feat/v020-config (currently open PR).

After each model response, a muted stats line is shown:

↑ 1.2k  ↓ 384  ·  45 tok/s  ·  2.3s

Implementation

• on_prediction_completed callback captures PredictionResult.stats per round
• _format_stats(stats_list, total_seconds) aggregates across multi-round turns
• Fields: prompt_tokens_count, predicted_tokens_count, tokens_per_second
• Falls back gracefully (empty string) if model doesn't expose stats

Closes #17

[VforVitorio]

Implemented in feat/ui-status-bar. After each model reply, a right-aligned stats line shows token count and generation speed (tokens/s). Controlled by /stats toggle and the show_stats UISettings field.