feat(server): hot-reload config store and reloadable infra clients

- Replace single Load() with a watching Store that re-reads config files
  (and config.local.yaml) and fans out a ReloadEvent with a per-field diff
  so consumers can decide whether the change is hot-applicable or requires
  a process restart.
- Wrap llm, retrieval, vector store, and object storage clients in
  Reloadable* shells so the bootstrap can swap their underlying impls when
  config changes without re-instantiating handlers.
- Make jwt.Manager and ops TokenIssuer mutable under a lock so secrets and
  TTLs can be rotated live; thread default plan code through a setter on
  the ops AdminUserService.
- Wire ConfigStore through bootstrap and every cmd/main.go, scheduler /
  worker / tenant-api / ops-api start the watcher; services and handlers
  take a config.Provider so they always read current values for things
  like generation.stream_enabled, scheduler dispatch, retrieval, etc.
- Switch shared/config decoding off viper to a Kratos-derived runtime
  package so env placeholders (\${VAR:default}) resolve consistently and
  the same source machinery powers both the loader and the watcher.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

server/cmd/worker-generate/main.go

@@ -25,7 +25,8 @@ func main() {
 	}
 	defer app.Close()
 
-	generationCfg := app.Config.Generation
+	generationProvider := tenantapp.NewGenerationStreamDisabledConfigProvider(app.ConfigStore)
+	generationCfg := app.Config().Generation
 	generationCfg.StreamEnabled = false
 
 	knowledgeSvc := tenantapp.NewKnowledgeService(
@@ -35,11 +36,11 @@ func main() {
 		app.ObjectStorage,
 		app.LLM,
 		app.Logger,
-		app.Config.Retrieval,
-		app.Config.LLM,
+		app.Config().Retrieval,
+		app.Config().LLM,
 		0,
 		0,
-	)
+	).WithConfigProvider(app.ConfigStore)
 
 	templateSvc := tenantapp.NewTemplateService(
 		app.DB,
@@ -52,8 +53,8 @@ func main() {
 		knowledgeSvc,
 		app.GenerationStreams,
 		generationCfg,
-		app.Config.LLM.MaxOutputTokens,
-	).WithCache(app.Cache)
+		app.Config().LLM.MaxOutputTokens,
+	).WithCache(app.Cache).WithConfigProvider(generationProvider)
 
 	promptRuleSvc := tenantapp.NewPromptRuleGenerationService(
 		app.DB,
@@ -62,8 +63,8 @@ func main() {
 		knowledgeSvc,
 		app.GenerationStreams,
 		generationCfg,
-		app.Config.LLM.MaxOutputTokens,
-	).WithCache(app.Cache).WithLogger(app.Logger)
+		app.Config().LLM.MaxOutputTokens,
+	).WithCache(app.Cache).WithLogger(app.Logger).WithConfigProvider(generationProvider)
 
 	publishJobSvc := tenantapp.NewPublishJobService(
 		app.DB,
@@ -80,8 +81,8 @@ func main() {
 		knowledgeSvc,
 		app.GenerationStreams,
 		generationCfg,
-		app.Config.LLM.MaxOutputTokens,
-	).WithCache(app.Cache)
+		app.Config().LLM.MaxOutputTokens,
+	).WithCache(app.Cache).WithConfigProvider(generationProvider)
 
 	kolWorker := generateworker.NewKolGenerationWorker(
 		app.DB,
@@ -89,12 +90,14 @@ func main() {
 		knowledgeSvc,
 		app.Logger,
 		generationCfg,
-		app.Config.LLM.MaxOutputTokens,
-	).WithCache(app.Cache)
+		app.Config().LLM.MaxOutputTokens,
+	).WithCache(app.Cache).WithConfigProvider(app.ConfigStore)
 
 	ctx, cancel := context.WithCancel(context.Background())
 	defer cancel()
 
+	app.StartConfigWatcher(ctx)
+
 	generateworker.NewArticleGenerationWorker(
 		app.DB,
 		app.RabbitMQ,
@@ -103,16 +106,16 @@ func main() {
 		imitationSvc,
 		kolWorker,
 		app.Logger,
-		app.Config.Generation,
-	).Start(ctx)
+		app.Config().Generation,
+	).WithConfigProvider(app.ConfigStore).Start(ctx)
 
 	generateworker.NewTemplateAssistWorker(
 		app.RabbitMQ,
 		templateSvc,
 		app.Logger,
-		app.Config.Generation.WorkerConcurrency,
-		app.Config.Generation.ArticleTimeout,
-	).Start(ctx)
+		app.Config().Generation.WorkerConcurrency,
+		app.Config().Generation.ArticleTimeout,
+	).WithConfigProvider(app.ConfigStore).Start(ctx)
 
 	app.Logger.Info("worker-generate started")