feat: Phase 2 - reliability & observability

- Circuit breaker: proper thresholds (3 failures, 30s timeout)
- HTTP timeouts: 30s on all providers (was no timeout)
- Structured logging: slog replaces fmt.Printf throughout
- Stream errors: propagated as SSE error events to client
- Registry fetch: retry with backoff (3 attempts)
- Registry reads in dashboard protected by RWMutex

internal/server/dashboard.go

@@ -14,6 +14,7 @@ import (
 	"gophergate/internal/db"
 	"gophergate/internal/models"
 	"gophergate/internal/utils"
+	"log/slog"
 
 	"github.com/shirou/gopsutil/v3/cpu"
 	"github.com/shirou/gopsutil/v3/disk"
@@ -879,6 +880,7 @@ func (s *Server) handleGetProviders(c *gin.Context) {
 
 		// Get models for this provider from registry
 		var models []string
+		s.registryMu.RLock()
 		if s.registry != nil {
 			registryID := id
 			if id == "gemini" {
@@ -897,6 +899,7 @@ func (s *Server) handleGetProviders(c *gin.Context) {
 				}
 			}
 		}
+		s.registryMu.RUnlock()
 
 		// If it's ollama, also include models from config
 		if id == "ollama" {
@@ -1060,6 +1063,7 @@ func (s *Server) handleGetModels(c *gin.Context) {
 	}
 
 	var result []gin.H
+	s.registryMu.RLock()
 	if s.registry != nil {
 		for pID, pInfo := range s.registry.Providers {
 			proxyProvider, allowed := allowedRegistryProviders[pID]
@@ -1210,6 +1214,7 @@ func (s *Server) handleUpdateModel(c *gin.Context) {
 
 	// Find provider for this model
 	providerID := "unknown"
+	s.registryMu.RLock()
 	if s.registry != nil {
 		for pID, pInfo := range s.registry.Providers {
 			if _, ok := pInfo.Models[id]; ok {
@@ -1388,6 +1393,7 @@ func (s *Server) handleSystemMetrics(c *gin.Context) {
 func (s *Server) handleGetSettings(c *gin.Context) {
 	providerCount := 0
 	modelCount := 0
+	s.registryMu.RLock()
 	if s.registry != nil {
 		providerCount = len(s.registry.Providers)
 		for _, p := range s.registry.Providers {