implement TrustScoreMetric for evaluating RAG source trustworthiness

[Danish2op]

Summary

Implements TrustScoreMetric as proposed in #2586.

What this adds

A new deterministic, rule-based metric that evaluates how trustworthy an
LLM's RAG output is based on the source tier of its retrieval context.

Scoring logic

Tier	Score
T1 (e.g. SEC filings)	1.0
T2 (e.g. Bloomberg)	0.8
T3 (e.g. news sites)	0.6
T4 (e.g. blog posts)	0.4
T5 (e.g. forums/AI)	0.2
Unmatched	0.5

Final score = average across all retrieval context chunks.

Usage

from deepeval.metrics import TrustScoreMetric
from deepeval.test_case import LLMTestCase

metric = TrustScoreMetric(
    threshold=0.7,
    source_tiers={"SEC filings": 1, "news": 3, "forums": 4}
)
test_case = LLMTestCase(
    input="What was Q3 revenue?",
    actual_output="Revenue was $4.2B",
    retrieval_context=["SEC 10-Q filing: Revenue $4.2B"]
)
metric.measure(test_case)
print(metric.score)    # 1.0
print(metric.reason)   # "Matched source 'SEC filings' mapped to Tier 1..."
print(metric.success)  # True

Testing

poetry run pytest tests/test_metrics/test_trust_score_metric.py

Closes #2586

[vercel]

@Danish2op is attempting to deploy a commit to the Confident AI Team on Vercel.

A member of the Team first needs to authorize it.