Experiment results URL points to nonexistent Kibana /runs/ route

At SDK commit `f86f4a013778c08ccbeb64195464f3378f13bbd5`, `log_results_url()` prints:

```text
{KIBANA_URL}/app/management/ai/evals/runs/{run_id}
```

Kibana 9.5.3 has no `/runs/:runId` evals route. Its detail route is `/experiments/:experimentId`. During a successful 30-example run, the SDK printed the `/runs/` link above; opening `/experiments/{RanExperiment.id}` instead showed all per-example results and evaluator statistics.

Sources:

- [SDK URL builder](https://github.com/elastic/elastic-evals-sdk-python/blob/f86f4a013778c08ccbeb64195464f3378f13bbd5/src/elastic_evals/utils/logging.py#L95)
- [Kibana 9.5.3 routes](https://github.com/elastic/kibana/blob/v9.5.3/x-pack/platform/plugins/shared/evals/public/application.tsx)

This is Python-SDK-maturity/UX: successful execution should return a usable experiment URL (ideally on `RanExperiment`) and print that same URL. Preserve any configured base/space path and strip userinfo credentials. Be explicit about execution grouping: the SDK's run ID and `RanExperiment.id` are different IDs, and simply replacing `/runs/` with `/experiments/` while retaining `run_id` is not sufficient without the appropriate execution query parameter.

Start with src/elastic_evals/utils/logging.py at the SDK URL builder and inspect how RanExperiment and the execution IDs are available. Compare the builder with Kibana's application.tsx routes and verify the configured base/space path and credential handling. Done means successful execution returns and prints a usable /experiments/{experimentId} URL with the appropriate execution grouping.

Written by the indexing model from the issue text.