feat: EU AI Act Compliance - Accuracy Metrics Reporting

[SergUdo]

Add Accuracy Metrics Reporting for EU AI Act Article 11

Why

EU AI Act (Article 11) requires transparency about system accuracy,
precision, and known limitations. Currently, ai-slop-gate provides
findings but doesn't report detection accuracy metrics.

Users need to understand:

• When to trust AI analysis results
• False positive rates per provider
• Confidence thresholds for different scenarios

What

Add accuracy metrics tracking and reporting to existing LLM providers
(Gemini, Groq, Static).

Benefits

• EU AI Act Article 11 compliance ready
• Users understand system reliability
• Better trust in results
• Professional transparency (not black box)
• Differentiates from competitors

TODO

• Add AccuracyMetrics class to track detection rates
• Implement confidence scoring per finding (0.0-1.0)
• Add accuracy statistics to JSON/HTML reports
• Document known limitations per provider in README
• Add --show-metrics flag to display accuracy stats

Example Output

{
  "accuracy_metrics": {
    "provider": "gemini",
    "hallucination_detection_rate": "94%",
    "false_positive_rate": "5%",
    "confidence_threshold": 0.7,
    "tested_scenarios": [
      "Python code: 95% accuracy",
      "TypeScript: 92% accuracy",
      "Complex logic: 85% accuracy"
    ]
  }
}

Labels: commercial-potential
Priority: Medium (EU AI Act enforcement 2026)
Complexity: Low (extend existing reports)

Docs:

• https://artificialintelligenceact.eu/article/11/
• EU AI Act Technical Documentation requirements