feat: Add comprehensive load balancing system documentation

[GdAyo19]

Closes #618

Add detailed module-level comments across the load balancing subsystem:

Load Prediction (time-series models):

• capacityPrediction.ts: Document hybrid forecasting algorithm (exponential smoothing + OLS regression + seasonality + CI widening)
• Add pipeline flow documentation for the full prediction lifecycle

Dynamic Load Distribution:

• rateLimiter.js: Document token bucket algorithm, priority queuing, endpoint-aware limiting, tier-based multipliers, and throttle modes
• api/middleware/rateLimiter.js: Document server-side sliding window
• CircuitBreaker.ts: Document CLOSED/OPEN/HALF_OPEN state machine and per-service breaker registry
• RetryManager.ts: Document exponential backoff with jitter and retryable error classification

Performance Monitoring:

• performanceMonitoring.js: Document Web Vitals, custom metrics, budget tracking, and the feedback loop into rate limiting/caching
• networkMonitoring.js: Document congestion calculation, fee prediction, validator monitoring, and throughput metrics

Infrastructure Integration:

• cacheManager.ts: Document 3-layer cache distribution strategy with write-through, SWR, and predictive prefetch
• cacheWarmingStrategy.ts: Document predictive warming scenarios driven by load predictions
• cacheAnalytics.ts: Document metrics collection for load balancer tuning
• scoringEngine.js: Document ML anomaly detection integration with RateLimiter and CircuitBreaker
• feePredictor.ts: Document fee prediction integration with load distribution
• SelfHealingManager.ts: Document auto-recovery for distribution components

[vercel]

@GdAyo19 is attempting to deploy a commit to the nanle-code's projects Team on Vercel.

A member of the Team first needs to authorize it.