New community metric: RAIL Score — responsible AI evaluation across 8 dimensions

## What

[RAIL Score](https://huggingface.co/spaces/responsible-ai-labs/rail_score) is a responsible AI evaluation metric now available as a community metric on the Hub. It scores LLM outputs across **8 dimensions** on a 0–10 scale:

| Dimension | What It Measures |
|-----------|-----------------|
| Fairness | Equitable treatment across demographic groups |
| Safety | Prevention of harmful or unsafe content |
| Reliability | Factual accuracy, internal consistency |
| Transparency | Clear communication of limitations and reasoning |
| Privacy | Protection of personal and sensitive data |
| Accountability | Traceability and auditability of decisions |
| Inclusivity | Accessible, inclusive language |
| User Impact | Positive value delivered to the user |

## Usage

```python
import evaluate

rail_score = evaluate.load("responsible-ai-labs/rail_score")

results = rail_score.compute(
predictions=["The capital of France is Paris."],
references=["What is the capital of France?"],
)

print(results["overall_score"]) # 8.2
print(results["overall_confidence"]) # 0.85
print(results["safety"]) # 9.5
print(results["fairness"]) # 8.0
```

## Features

- All 8 RAIL dimensions with per-dimension and per-example scores
- **Confidence scores** for every dimension and overall
- **Custom dimension weights** (e.g., prioritize safety over inclusivity)
- `basic` and `deep` evaluation modes (deep includes grounded explanations)
- Domain-specific scoring: `general`, `healthcare`, `finance`, `legal`, `education`, `code`
- Issue detection with dimension-level flagging

## Deep mode example

```python
results = rail_score.compute(
predictions=responses,
references=prompts,
mode="deep",
include_explanations=True,
include_issues=True,
weights={"safety": 25, "reliability": 20, "fairness": 15,
"transparency": 10, "privacy": 10, "accountability": 5,
"inclusivity": 10, "user_impact": 5},
)

# Per-dimension explanations
for dim, explanations in results["explanations"].items():
print(f"{dim}: {explanations[0]}")
```

## Links

- **Hub Space**: [responsible-ai-labs/rail_score](https://huggingface.co/spaces/responsible-ai-labs/rail_score)
- **SDK on PyPI**: [rail-score-sdk](https://pypi.org/project/rail-score-sdk/)
- **Documentation**: [docs.responsibleailabs.ai](https://docs.responsibleailabs.ai)

Requires a RAIL Score API key (free tier available at [responsibleailabs.ai](https://responsibleailabs.ai)).

Would be great to get this listed alongside other community evaluation metrics. Happy to make any changes needed for better integration with the evaluate ecosystem.