LLM Output Consistency Checker

Implement self-consistency checking for LLM outputs.

## Task Type
{{task_type}}

## Consistency Requirements
{{consistency_requirements}}

## Budget for Multiple Calls
{{budget_constraints}}

Build consistency checking:

```python
class ConsistencyChecker:
    def generate_variants(self, prompt: str, n: int, temperature: float) -> List[str]:
        """Generate multiple response variants"""
        pass
    
    def check_consistency(self, responses: List[str]) -> ConsistencyResult:
        """
        Analyze:
        - Agreement rate
        - Contradiction detection
        - Uncertainty estimation
        """
        pass
    
    def select_consensus(self, responses: List[str], method: str) -> str:
        """
        Methods: majority_vote, weighted_vote, merge
        """
        pass
```

Include:
- Temperature strategies
- Prompt variation techniques
- Cost-quality trade-off analysis
- Async parallel generation