🧠 Implement AI-Driven Intelligent Investigation System

[Copilot]

This PR introduces a comprehensive AI-driven investigation system that transforms AppInsights Detective from a query tool into an intelligent problem-solving platform. The system automatically analyzes Application Insights problems, generates dynamic investigation plans, and provides root cause analysis with actionable recommendations.

Key Features

🗣️ Natural Language Problem Analysis

Users can now describe problems in plain English, and the AI will automatically classify them into investigation types:

aidx investigate "Application is responding slowly"
aidx investigate "Users getting 500 errors since 2 PM"
aidx investigate "Missing telemetry data for mobile app"

🧠 Intelligent Investigation Planning

The system generates multi-phase investigation plans tailored to problem types:

• Performance: Response time analysis, dependency performance, resource utilization
• Availability: Error rate analysis, downtime timeline, service health checks
• Data Quality: Missing telemetry detection, consistency validation, sampling verification
• Dependencies: External service analysis, connection patterns, timeout investigation

🔍 Automated Execution & Analysis

Each investigation executes systematically through multiple phases:

1. Problem Classification with AI confidence scoring
2. Dynamic Plan Generation with specific KQL queries for each phase
3. Systematic Execution with evidence collection and significance scoring
4. Root Cause Analysis with confidence-based primary cause identification
5. Actionable Recommendations (immediate, short-term, long-term actions)

💻 Complete CLI Integration

The new investigate command provides comprehensive investigation management:

# Interactive guided investigation
aidx investigate --interactive

# Investigation management
aidx investigate --status <id>
aidx investigate --history
aidx investigate --pause <id>
aidx investigate --resume <id>

# Export results in multiple formats
aidx investigate --export <id> --format markdown

Example Investigation Flow

$ aidx investigate "Application response times are very slow" --interactive

🔍 Starting AI-Driven Investigation
Problem: Application response times are very slow
Type: performance (confidence: 92%)

📋 Investigation Plan (3 phases, ~4 minutes)
🔄 Executing Investigation...
📊 Progress: ████████████████████ 100%

✅ Investigation Completed!
Root Cause: Database connection pool exhaustion
Confidence: 87.5%
Evidence: 12 items collected

Key Evidence:
• Database connection pool utilization peaked at 98%
• Response time P95 increased from 245ms to 1.2s
• Connection timeouts increased 450% during incident

Technical Implementation

Architecture Integration

• Seamless Integration: Uses existing AI providers, data sources, and authentication
• Service Layer: New IntelligentInvestigationService with comprehensive investigation orchestration
• DI Container: Properly registered in Bootstrap with full dependency injection
• Session Management: Integrates with existing session system for context tracking

AI-Powered Analysis

• Problem Classification: Advanced prompt engineering for accurate problem type detection
• Plan Generation: Investigation type-specific templates for comprehensive analysis
• Result Analysis: AI-powered evidence significance assessment and root cause identification

Robust Testing

• Service Tests: 21 comprehensive tests covering all major functionality
• CLI Tests: 7 tests covering command structure and integration
• Mock Infrastructure: Complete testing framework with AI, data source, and session mocks
• Error Handling: Comprehensive testing of failure scenarios and graceful degradation

Benefits

For Users

• ⏱️ Time Reduction: Investigation time reduced from hours to minutes
• 🎓 Lower Barrier: No deep KQL expertise required for comprehensive analysis
• 🔍 Systematic Approach: Evidence-based investigation with AI-powered insights
• 💡 Actionable Results: Clear immediate and long-term recommendations

For Teams

• 📚 Knowledge Sharing: Consistent investigation methodologies across teams
• 🎯 Focus Shift: From manual diagnosis to solution implementation
• 📊 Effectiveness Tracking: Investigation history and success metrics

Implementation Statistics

• Files Added/Modified: 15 files (11 new, 4 modified)
• Lines of Code: ~2,475 lines of production code
• Test Coverage: 29 new tests (422 total tests, 100% passing)
• Documentation: Complete user guide and CLI integration

The system maintains full backward compatibility while adding powerful new investigation capabilities that make Application Insights analysis accessible to users of all skill levels.

Fixes #122.

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.