server: expose speculative decoding counters in Prometheus metrics

[boxcee]

Adds two new counters to the /metrics Prometheus endpoint:

• llamacpp:spec_tokens_drafted_total
• llamacpp:spec_tokens_accepted_total

Divide accepted by drafted to get acceptance rate.

The counters reuse n_draft_total / n_draft_accepted introduced in #22673. Those fields are already tracked per slot during speculative decoding. No new tracking logic, just plumbing those values to the metrics handler via server_metrics::on_prediction().

Without this, the only way to watch acceptance rate is to grep the server logs. Grafana and similar tools can now track it directly.

Changes:

• server_metrics: two new uint64_t fields, incremented in on_prediction()
• server_task_result_metrics: two new fields to carry the values to the HTTP handler
• Prometheus handler: two new counter entries in all_metrics_def
• README: metrics table updated

---

AI disclosure: I used Claude Code to help locate the relevant code paths and scaffold the initial implementation. I've reviewed every line and can explain it.

[ggml-gh-bot]

Hi @boxcee, thanks for your contribution!

Per our contribution guidelines, the automated PR checker found the following issue(s) that need your attention:

• AI-generated content: This project does not accept PRs, descriptions or commit messages that are fully or predominantly AI-generated. If you have used AI to assist you in writing code, please make sure to disclose that explicitly.

---

Please note that maintainers reserve the right to make final decisions on PRs. If you believe there is a mistake, please comment below.