fix(model): DeepSeek 推理参数语义反转 + thinking 开关真发（按官方文档核对结果）

依据 DevDocs/design/04 §3.2「官方核对后的修正」9 条 + 本地 DevDocs/reference/deepseek/facts-for-cyrene.md（官方原句/URL）。
- 🔴 reasoning_effort 白名单改为官方集合 [none, low, high, max] + 别名映射(minimal→low, medium→high, xhigh→high, ultra→max)；**删除就近降级**，非法值一律丢弃+WARN。此前 none(关思考) 被降级成 low(开思考) 属语义反转
- 🔴 DeepSeek thinking 开关真发顶层 {"thinking":{"type":"enabled|disabled"}}（官方默认 enabled，不传≠关）；此前只打 WARN 导致 mode:off 静默无效
- DeepSeek 的 /anthropic 端点忽略 budget_tokens → 不再对其抬 max_tokens（会误导预算/日志），改为按官方用 output_config.effort
- 注释纠正：penalty 属接口级 deprecated（非仅思考模式）；top_p 仅思考模式生效且下限 0.95（不加该字段）
- tool_choice 雷登记：思考模式下 required/指定 tool 会 400（思考默认开）
- 旧模型名清理（deepseek-chat/reasoner 已于 2026-07-24 停用）
- 请求体级测试 + 变异验证（把 none 改回 low / 把 thinking 改回只 WARN → 对应断言 FAIL）

internal/services/gateway/admin_write_test.go

@@ -231,7 +231,7 @@ func TestAdminAddLlmProviderPersistsAndReloads(t *testing.T) {
 	tok := signAny(t, jwt, "acct", auth.RoleOwner)
 
 	resp, body := doReq(t, http.MethodPost, ts.URL+"/api/v1/admin/llm",
-		map[string]any{"id": "deepseek", "provider": "openai", "baseUrl": "https://api.deepseek.com/v1", "apiKey": "sk-ds", "model": "deepseek-chat", "multimodal": true, "maxTokens": 1234}, tok)
+		map[string]any{"id": "deepseek", "provider": "openai", "baseUrl": "https://api.deepseek.com/v1", "apiKey": "sk-ds", "model": "deepseek-flash", "multimodal": true, "maxTokens": 1234}, tok)
 	if resp.StatusCode != http.StatusCreated {
 		t.Fatalf("status = %d, want 201 (body=%s)", resp.StatusCode, body)
 	}
@@ -243,7 +243,7 @@ func TestAdminAddLlmProviderPersistsAndReloads(t *testing.T) {
 	if err := json.Unmarshal(decodeBody(t, body)["data"], &res); err != nil {
 		t.Fatalf("decode add: %v", err)
 	}
-	if !res.Persisted || res.Path != modelsPath || res.Provider.Model != "deepseek-chat" || res.Provider.MaxTokens != 1234 {
+	if !res.Persisted || res.Path != modelsPath || res.Provider.Model != "deepseek-flash" || res.Provider.MaxTokens != 1234 {
 		t.Fatalf("add res = %+v", res)
 	}
 
@@ -257,8 +257,8 @@ func TestAdminAddLlmProviderPersistsAndReloads(t *testing.T) {
 	}
 
 	// The running stack reflects the new active provider/model (hot reload).
-	if runtime.Provider() != "openai" || runtime.Model() != "deepseek-chat" {
-		t.Fatalf("runtime Provider=%q Model=%q, want openai/deepseek-chat", runtime.Provider(), runtime.Model())
+	if runtime.Provider() != "openai" || runtime.Model() != "deepseek-flash" {
+		t.Fatalf("runtime Provider=%q Model=%q, want openai/deepseek-flash", runtime.Provider(), runtime.Model())
 	}
 
 	// GET /admin/llm reflects the saved group and never leaks any API key.
@@ -270,8 +270,8 @@ func TestAdminAddLlmProviderPersistsAndReloads(t *testing.T) {
 	if err := json.Unmarshal(decodeBody(t, gBody)["data"], &list); err != nil {
 		t.Fatalf("decode list: %v", err)
 	}
-	if len(list.Providers) == 0 || list.Providers[0].Model != "deepseek-chat" || list.Providers[0].MaxTokens != 1234 {
-		t.Fatalf("list = %+v, want deepseek-chat first with maxTokens 1234", list.Providers)
+	if len(list.Providers) == 0 || list.Providers[0].Model != "deepseek-flash" || list.Providers[0].MaxTokens != 1234 {
+		t.Fatalf("list = %+v, want deepseek-flash first with maxTokens 1234", list.Providers)
 	}
 	if strings.Contains(string(gBody), "sk-ds") || strings.Contains(string(gBody), "sk-env") {
 		t.Fatalf("API key leaked in GET /admin/llm: %s", gBody)
@@ -311,7 +311,7 @@ func TestAdminAddLlmProviderPartialEditPreservesFields(t *testing.T) {
 
 	// Seed a fully-specified group.
 	resp, body := doReq(t, http.MethodPost, ts.URL+"/api/v1/admin/llm",
-		map[string]any{"id": "deepseek", "provider": "openai", "baseUrl": "https://api.deepseek.com/v1", "apiKey": "sk-ds", "model": "deepseek-chat", "multimodal": true, "fallback": "deepseek-lite", "maxTokens": 1234}, tok)
+		map[string]any{"id": "deepseek", "provider": "openai", "baseUrl": "https://api.deepseek.com/v1", "apiKey": "sk-ds", "model": "deepseek-flash", "multimodal": true, "fallback": "deepseek-v4-flash", "maxTokens": 1234}, tok)
 	if resp.StatusCode != http.StatusCreated {
 		t.Fatalf("seed status = %d, want 201 (body=%s)", resp.StatusCode, body)
 	}
@@ -320,7 +320,7 @@ func TestAdminAddLlmProviderPartialEditPreservesFields(t *testing.T) {
 	// stored value — in particular maxTokens (a full replace reset it to 0) and
 	// apiKey (a full replace wiped it).
 	resp2, body2 := doReq(t, http.MethodPost, ts.URL+"/api/v1/admin/llm",
-		map[string]any{"id": "deepseek", "model": "deepseek-chat-v2"}, tok)
+		map[string]any{"id": "deepseek", "model": "deepseek-v4-pro"}, tok)
 	if resp2.StatusCode != http.StatusCreated {
 		t.Fatalf("edit status = %d, want 201 (body=%s)", resp2.StatusCode, body2)
 	}
@@ -331,7 +331,7 @@ func TestAdminAddLlmProviderPartialEditPreservesFields(t *testing.T) {
 		t.Fatalf("decode edit: %v", err)
 	}
 	p := res.Provider
-	if p.Model != "deepseek-chat-v2" || p.MaxTokens != 1234 || p.Fallback != "deepseek-lite" ||
+	if p.Model != "deepseek-v4-pro" || p.MaxTokens != 1234 || p.Fallback != "deepseek-v4-flash" ||
 		!p.Multimodal || p.Provider != "openai" || p.BaseURL != "https://api.deepseek.com/v1" {
 		t.Fatalf("edit provider = %+v, want model updated and every omitted field preserved", p)
 	}
@@ -340,7 +340,7 @@ func TestAdminAddLlmProviderPartialEditPreservesFields(t *testing.T) {
 	if warn != nil {
 		t.Fatalf("load models warn = %v", warn)
 	}
-	if len(groups) == 0 || groups[0].APIKey != "sk-ds" || groups[0].MaxTokens != 1234 || groups[0].Model != "deepseek-chat-v2" {
+	if len(groups) == 0 || groups[0].APIKey != "sk-ds" || groups[0].MaxTokens != 1234 || groups[0].Model != "deepseek-v4-pro" {
 		t.Fatalf("file groups = %+v, want apiKey/maxTokens preserved and model updated", groups)
 	}
 }