[FEATURE] Implement ADR quality checker with LLM integration

## ✨ Feature Request Template

### **Feature Description**
Implement an ADR quality checker that uses LLM capabilities to analyze existing ADR files and provide improvement suggestions, helping teams maintain high-quality architectural documentation.

### **Use Case**
Teams need to ensure their ADRs are well-written and complete:
- New team members may not know what makes a good ADR
- Existing ADRs may be missing important sections or details
- Quality varies across different authors and time periods
- Manual review is time-consuming and inconsistent
- Teams want to maintain documentation standards automatically
- Continuous improvement of architectural decision documentation

### **Proposed Solution**
Implement a quality checker that:
- Analyzes existing ADR files for completeness and clarity
- Uses LLM to evaluate content quality against best practices
- Provides specific, actionable improvement suggestions
- Operates in read-only mode (no file modifications)
- Integrates cleanly with existing service layer architecture
- Supports batch checking of multiple ADR files
- Provides both summary and detailed feedback modes

Command interface: `adr-ai-tools check ` or `adr-ai-tools check docs/adr/`

### **Alternatives Considered**
1. **Rule-based checker**: Fast but limited to structural checks, misses content quality
2. **Template validation only**: Simple but doesn't assess actual content quality
3. **Manual review process**: Thorough but doesn't scale, inconsistent results
4. **External linting tools**: Exist for code but not specialized for ADR content
5. **LLM-powered analysis**: More sophisticated, can assess content quality and clarity

### **Acceptance Criteria**
Define what "done" looks like for this feature:
- [ ] ADR documented for LLM integration strategy and quality criteria
- [ ] LLM service implemented following service layer architecture (ADR-0015)
- [ ] CLI command `adr-ai-tools check ` analyzes single ADR file
- [ ] CLI command `adr-ai-tools check ` analyzes multiple ADR files
- [ ] Quality assessment covers: completeness, clarity, structure, rationale quality
- [ ] Specific, actionable improvement suggestions provided
- [ ] Configuration integration for LLM provider selection and API keys
- [ ] Graceful handling of API failures and rate limits
- [ ] Read-only operation with no file modifications
- [ ] Unit tests for LLM service with mocked responses
- [ ] E2E tests with sample ADR files
- [ ] Documentation for setup and usage

### **Additional Context**
This is the first LLM integration feature and will establish patterns for future AI-powered capabilities. Key considerations:

**Technical Integration:**
- Builds on existing FileSystemService for reading ADR files
- Requires new LLMService following established service patterns
- Should integrate with configuration system (Issue #20) for API key management
- Error handling for network issues, API limits, invalid responses

**Quality Assessment Areas:**
- Structural completeness (all required sections present)
- Content clarity and readability
- Rationale strength and completeness
- Alternative consideration thoroughness
- Future implications coverage

**Future Expansion:**
This foundation will enable other LLM features like title generation (suggested in planning), content drafting assistance, and ADR summarization capabilities.

The implementation should be designed to easily support multiple LLM providers (OpenAI, Anthropic, local models) through a common interface.