fix: 三端联调测试 2 P0 + 6 P1 + 2 P2 修复

P0-1: SaaS relay 模型别名解析 — "glm-4-flash" → "glm-4-flash-250414" (resolve_model)
P0-2: config.rs interpolate_env_vars UTF-8 修复 (chars 迭代器替代 bytes as char)
      + DB 启动编码检查 + docker-compose UTF-8 编码参数

P1-3: UI 模型选择器覆盖 Agent 默认模型 (model_override 全链路: TS→Tauri→Rust kernel)
P1-6: 知识搜索管道修复 — seed_knowledge 创建 chunks + 默认分类 (seed/uploaded/distillation)
P1-7: 用量限额从当前 Plan 读取 (非 stale usage 表)
P1-8: relay 双维度配额检查 (relay_requests + input_tokens)

P2-9: SSE 路径 token 计数修复 — 流结束检测替代固定 500ms sleep + billing increment

crates/zclaw-kernel/src/kernel/messaging.rs

@@ -25,7 +25,7 @@ impl Kernel {
         agent_id: &AgentId,
         message: String,
     ) -> Result<MessageResponse> {
-        self.send_message_with_chat_mode(agent_id, message, None).await
+        self.send_message_with_chat_mode(agent_id, message, None, None).await
     }
 
     /// Send a message to an agent with optional chat mode configuration
@@ -34,6 +34,7 @@ impl Kernel {
         agent_id: &AgentId,
         message: String,
         chat_mode: Option<ChatModeConfig>,
+        model_override: Option<String>,
     ) -> Result<MessageResponse> {
         let agent_config = self.registry.get(agent_id)
             .ok_or_else(|| zclaw_types::ZclawError::NotFound(format!("Agent not found: {}", agent_id)))?;
@@ -41,12 +42,16 @@ impl Kernel {
         // Create or get session
         let session_id = self.memory.create_session(agent_id).await?;
 
-        // Use agent-level model if configured, otherwise fall back to global config
-        let model = if !agent_config.model.model.is_empty() {
-            agent_config.model.model.clone()
-        } else {
-            self.config.model().to_string()
-        };
+        // Model priority: UI override > Agent config > Global config
+        let model = model_override
+            .filter(|m| !m.is_empty())
+            .unwrap_or_else(|| {
+                if !agent_config.model.model.is_empty() {
+                    agent_config.model.model.clone()
+                } else {
+                    self.config.model().to_string()
+                }
+            });
 
         // Create agent loop with model configuration
         let subagent_enabled = chat_mode.as_ref().and_then(|m| m.subagent_enabled).unwrap_or(false);
@@ -122,7 +127,7 @@ impl Kernel {
         agent_id: &AgentId,
         message: String,
     ) -> Result<mpsc::Receiver<zclaw_runtime::LoopEvent>> {
-        self.send_message_stream_with_prompt(agent_id, message, None, None, None).await
+        self.send_message_stream_with_prompt(agent_id, message, None, None, None, None).await
     }
 
     /// Send a message with streaming, optional system prompt, optional session reuse,
@@ -134,6 +139,7 @@ impl Kernel {
         system_prompt_override: Option<String>,
         session_id_override: Option<zclaw_types::SessionId>,
         chat_mode: Option<ChatModeConfig>,
+        model_override: Option<String>,
     ) -> Result<mpsc::Receiver<zclaw_runtime::LoopEvent>> {
         let agent_config = self.registry.get(agent_id)
             .ok_or_else(|| zclaw_types::ZclawError::NotFound(format!("Agent not found: {}", agent_id)))?;
@@ -150,12 +156,16 @@ impl Kernel {
             None => self.memory.create_session(agent_id).await?,
         };
 
-        // Use agent-level model if configured, otherwise fall back to global config
-        let model = if !agent_config.model.model.is_empty() {
-            agent_config.model.model.clone()
-        } else {
-            self.config.model().to_string()
-        };
+        // Model priority: UI override > Agent config > Global config
+        let model = model_override
+            .filter(|m| !m.is_empty())
+            .unwrap_or_else(|| {
+                if !agent_config.model.model.is_empty() {
+                    agent_config.model.model.clone()
+                } else {
+                    self.config.model().to_string()
+                }
+            });
 
         // Create agent loop with model configuration
         let subagent_enabled = chat_mode.as_ref().and_then(|m| m.subagent_enabled).unwrap_or(false);