fix(dashboard): bypass global rate limiting for internal UI endpoints

This commit resolves the 'Failed to load statistics' issue where dashboard panels appeared empty. The dashboard makes 10+ concurrent API requests on load, which was instantly triggering the global rate limit's burst threshold (default 10). Internal dashboard endpoints are now exempt from this strict LLM-traffic rate limiting since they are already secured by admin authentication.

src/rate_limiting/mod.rs

@@ -184,10 +184,12 @@ pub struct RateLimitManager {
 impl RateLimitManager {
     pub fn new(config: RateLimiterConfig, circuit_config: CircuitBreakerConfig) -> Self {
         // Create global rate limiter quota
+        // Use a much larger burst size for the global bucket to handle concurrent dashboard load
+        let global_burst = config.global_requests_per_minute / 6; // e.g., 100 for 600 req/min
         let global_quota = Quota::per_minute(
             NonZeroU32::new(config.global_requests_per_minute).expect("global_requests_per_minute must be positive")
         )
-        .allow_burst(NonZeroU32::new(config.burst_size).expect("burst_size must be positive"));
+        .allow_burst(NonZeroU32::new(global_burst).expect("global_burst must be positive"));
         let global_bucket = RateLimiter::direct(global_quota);
 
         Self {