feat(knowledge): multi-query retrieval with LLM query rewrite

Add a shared structured query builder (knowledge_query.go) and switch
all generation paths to it, so retrieval queries carry task intent
(type, brand, region, keywords, key points) instead of a raw prompt.

ResolveContext now rewrites the query into multiple sub-questions via
the URL-markdown model, embeds and searches each, and merges/dedupes
candidates before rerank; falls back to parsed fallback questions when
rewrite is unavailable or returns too few. Resolve logs gain query
list, count, and rewrite stage/model for observability.

- knowledge_query.go/_test.go: BuildKnowledgeQuery + intent normalization
- knowledge_service.go: per-query embed/search loop, query rewrite, logs
- template_prompt/template_service/prompt_generate/article_imitation/
  kol_generation_worker: adopt the shared query builder
- generation_observability: richer task error logging

server/internal/worker/generate/article_generation_worker.go

@@ -251,7 +251,7 @@ func (w *ArticleGenerationWorker) handleTaskFailure(ctx context.Context, task te
 	task.TaskType = strings.TrimSpace(task.TaskType)
 	logCtx := generationWorkerTaskLogContext(task, stage, "", "failed", tenantapp.GenerationTaskResultFailure, 0, "")
 	tenantapp.RecordGenerationTaskEvent(tenantapp.GenerationTaskEventFailure, logCtx)
-	tenantapp.LogGenerationTaskWarn(w.logger, "article generation task failed", logCtx, zap.Error(taskErr))
+	tenantapp.LogGenerationTaskWarn(w.logger, "article generation task failed", logCtx, tenantapp.GenerationTaskErrorFields(taskErr)...)
 	var err error
 	switch {
 	case task.TaskType == kolGenerationTaskType && w.kolWorker != nil: