feat: Production readiness improvements for WHOOSH council formation

Major security, observability, and configuration improvements:

## Security Hardening
- Implemented configurable CORS (no more wildcards)
- Added comprehensive auth middleware for admin endpoints
- Enhanced webhook HMAC validation
- Added input validation and rate limiting
- Security headers and CSP policies

## Configuration Management
- Made N8N webhook URL configurable (WHOOSH_N8N_BASE_URL)
- Replaced all hardcoded endpoints with environment variables
- Added feature flags for LLM vs heuristic composition
- Gitea fetch hardening with EAGER_FILTER and FULL_RESCAN options

## API Completeness
- Implemented GetCouncilComposition function
- Added GET /api/v1/councils/{id} endpoint
- Council artifacts API (POST/GET /api/v1/councils/{id}/artifacts)
- /admin/health/details endpoint with component status
- Database lookup for repository URLs (no hardcoded fallbacks)

## Observability & Performance
- Added OpenTelemetry distributed tracing with goal/pulse correlation
- Performance optimization database indexes
- Comprehensive health monitoring
- Enhanced logging and error handling

## Infrastructure
- Production-ready P2P discovery (replaces mock implementation)
- Removed unused Redis configuration
- Enhanced Docker Swarm integration
- Added migration files for performance indexes

## Code Quality
- Comprehensive input validation
- Graceful error handling and failsafe fallbacks
- Backwards compatibility maintained
- Following security best practices

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

internal/composer/models.go

@@ -189,6 +189,27 @@ type ComposerConfig struct {
 	AnalysisTimeoutSecs int  `json:"analysis_timeout_secs"`
 	EnableCaching       bool `json:"enable_caching"`
 	CacheTTLMins       int  `json:"cache_ttl_mins"`
+	
+	// Feature flags
+	FeatureFlags FeatureFlags `json:"feature_flags"`
+}
+
+// FeatureFlags controls experimental and optional features in the composer
+type FeatureFlags struct {
+	// LLM-based analysis (vs heuristic-based)
+	EnableLLMClassification   bool `json:"enable_llm_classification"`
+	EnableLLMSkillAnalysis    bool `json:"enable_llm_skill_analysis"`
+	EnableLLMTeamMatching     bool `json:"enable_llm_team_matching"`
+	
+	// Advanced analysis features  
+	EnableComplexityAnalysis  bool `json:"enable_complexity_analysis"`
+	EnableRiskAssessment      bool `json:"enable_risk_assessment"`
+	EnableAlternativeOptions  bool `json:"enable_alternative_options"`
+	
+	// Performance and debugging
+	EnableAnalysisLogging     bool `json:"enable_analysis_logging"`
+	EnablePerformanceMetrics  bool `json:"enable_performance_metrics"`
+	EnableFailsafeFallback    bool `json:"enable_failsafe_fallback"`
 }
 
 // DefaultComposerConfig returns sensible defaults for MVP
@@ -204,5 +225,26 @@ func DefaultComposerConfig() *ComposerConfig {
 		AnalysisTimeoutSecs: 60,
 		EnableCaching:       true,
 		CacheTTLMins:       30,
+		FeatureFlags: DefaultFeatureFlags(),
+	}
+}
+
+// DefaultFeatureFlags returns conservative defaults that prioritize reliability
+func DefaultFeatureFlags() FeatureFlags {
+	return FeatureFlags{
+		// LLM features disabled by default - use heuristics for reliability
+		EnableLLMClassification:   false,
+		EnableLLMSkillAnalysis:    false,
+		EnableLLMTeamMatching:     false,
+		
+		// Basic analysis features enabled
+		EnableComplexityAnalysis:  true,
+		EnableRiskAssessment:      true,
+		EnableAlternativeOptions:  false, // Disabled for MVP performance
+		
+		// Debug and monitoring enabled
+		EnableAnalysisLogging:     true,
+		EnablePerformanceMetrics:  true,
+		EnableFailsafeFallback:    true,
 	}
 }