fix(intelligence): 精确化 dead_code 标注并实现 LLM 上下文压缩

- 将 intelligence/llm/memory/browser 模块的 dead_code 注释从模糊的
  "reserved for future" 改为明确说明 Tauri invoke_handler 运行时注册机制
- 为 identity.rs 中 3 个真正未使用的方法添加 #[allow(dead_code)]
- 实现 compactor use_llm: true 功能：新增 compact_with_llm 方法和
  compactor_compact_llm Tauri 命令，支持 LLM 驱动的对话摘要生成
- 将 pipeline_commands.rs 中 40+ 处 println!/eprintln! 调试输出替换为
  tracing::debug!/warn!/error! 结构化日志
- 移除 intelligence/mod.rs 中不必要的 #[allow(unused_imports)]

desktop/src-tauri/src/llm/mod.rs

@@ -1,9 +1,11 @@
 //! LLM Client Module
 //!
-//! Provides LLM API integration for memory extraction.
+//! Provides LLM API integration for memory extraction and embedding.
 //! Supports multiple providers with a unified interface.
 //!
-//! Note: Some fields are reserved for future streaming and provider selection features
+//! NOTE: #[tauri::command] functions are registered via invoke_handler! at runtime,
+// which the Rust compiler does not track as "use". Module-level allow required
+// for Tauri-commanded functions and internal type definitions.
 
 #![allow(dead_code)]
 
@@ -357,6 +359,11 @@ impl EmbeddingClient {
         }
     }
 
+    /// Check if the embedding client is properly configured and available.
+    pub fn is_configured(&self) -> bool {
+        self.config.provider != "local" && !self.config.api_key.is_empty()
+    }
+
     pub async fn embed(&self, text: &str) -> Result<EmbeddingResponse, String> {
         if self.config.provider == "local" || self.config.api_key.is_empty() {
             return Err("Local TF-IDF mode does not support API embedding".to_string());