agent: 更智能的 LLM provider 调度（byModel 精确路由 + AUTO 优先级链）

吸收 llmsproxy 的调度思想适配 HomeAgent“一源一模型”结构：
- RoutableProvider{Model,Priority} 次级接口（不破坏既有 Provider 实现）
- ProviderManager.OrderedProviders 改为按 (优先级 desc, 可用, 默认优先) 稳定排序，
  AUTO/空模型走该优先级链
- 新增 ProviderManager.ResolveForModel：精确模型名路由到归属源，找不到回落 AUTO 链
- LuaAdaptedProvider 不再无条件覆写 req.Model；显式模型名原样转发
- LLMSource.Priority + core.llm.sources.<name>.priority 配置项
- process.go: 显式模型走 ResolveForModel，AUTO 走 OrderedProviders
- 新增路由单测（优先级排序 + byModel 解析）

验证: go test ./... 27 包 0 失败；Windows 交叉编译通过；部署后服务健康

internal/agent/core/process.go

@@ -70,7 +70,13 @@ func (a *Agent) process(input string, stageCtx *sdk.StageContext) (response stri
 
 		var providers []agentAPI.Provider
 		if a.providerManager != nil {
-			allProviders := a.providerManager.OrderedProviders()
+			// 精确模型名走 byModel 路由；AUTO/空走优先级链
+			var allProviders []agentAPI.Provider
+			if req.Model != "" && !strings.EqualFold(req.Model, "AUTO") {
+				allProviders = a.providerManager.ResolveForModel(req.Model)
+			} else {
+				allProviders = a.providerManager.OrderedProviders()
+			}
 			providers = make([]agentAPI.Provider, 0, len(allProviders))
 			for _, p := range allProviders {
 				if a.providerManager.IsAvailable(p.Name()) {
@@ -156,11 +162,11 @@ func (a *Agent) process(input string, stageCtx *sdk.StageContext) (response stri
 		resp.ToolCalls = convertBackToolCalls(stageCtx.ToolCalls)
 
 		chainPayload := map[string]interface{}{
-			"content":          resp.Content,
-			"reasoning":        resp.ReasoningContent,
-			"tool_calls":       resp.ToolCalls,
-			"phase":            "intermediate",
-			"turn":             turn,
+			"content":    resp.Content,
+			"reasoning":  resp.ReasoningContent,
+			"tool_calls": resp.ToolCalls,
+			"phase":      "intermediate",
+			"turn":       turn,
 		}
 		if resp.TokenUsage.Total > 0 {
 			chainPayload["usage"] = map[string]int{