feat(aibox): L-batch profile polish (display name, prefix escape, temperature, reasoning aliases)

L1 display name: client.name now resolves to the profile's
display_name field, falling back to the profile key — the dialogue
context no longer labels Kimi/Qwen sessions "DeepSeek". The legacy
flat api section keeps "DeepSeek" (zero breakage).
_resolve_model_profile now returns (profile, model, profile_name).

L2 prefix escape: "model:NAME" forces bare-model semantics even when
a profile happens to be named NAME; "profile:NAME" forces profile
semantics and raises a clear error (listing available profiles) when
missing, instead of silently falling back to bare-name behavior.
Empty prefixes are rejected. Wired through --model / create_llm_client
/ subagent model params.

L3 profile temperature: new profile field "temperature" becomes the
client's default_temperature. chat_completion/stream_chat treat
temperature=None as "use profile default, else the old hardcoded
0.3/0.7", so explicit caller values still win. DeepSeekChat and
run_light_subagent resolve CLI > profile > constants.DEFAULT_TEMPERATURE.

L5 reasoning field aliases: _parse_delta recognizes
reasoning_content / reasoning / thinking / thinking_content (first
non-empty wins), so non-DeepSeek vendors' thinking streams are no
longer silently dropped.

L4 (cross-profile failover) intentionally deferred: streaming
failover needs request-replay semantics and is feature-level work.

Tests: new TestLBatch (16 cases); 7 existing unpack sites updated to
the 3-tuple. Full suite 315 passed.

aibox/aibox.py

@@ -585,9 +585,12 @@ def run_light_subagent(role, task, config, logger=None, model=None,
     try:
         # 幂等校验：限制 max_tokens 范围，防止意外超限大量消耗
         max_tokens = max(1, min(int(max_tokens or 2048), constants.DEFAULT_MAX_TOKENS))
-        temp = temperature if temperature is not None else constants.DEFAULT_TEMPERATURE
         # 子代理使用独立的 LLM 客户端（可选不同模型），独立上下文，不污染主对话
         sub_client = create_llm_client(config, api_key=None, logger=logger, model=model)
+        # L3：调用方 temperature > 档案级 default_temperature > 全局默认
+        client_temp = getattr(sub_client, "default_temperature", None)
+        temp = temperature if temperature is not None else (
+            client_temp if client_temp is not None else constants.DEFAULT_TEMPERATURE)
         messages = [
             {"role": "system", "content": role},
             {"role": "user", "content": task}
@@ -915,7 +918,11 @@ class DeepSeekChat:
         self.api_key = self.client.api_key
         self.name = self.client.name
         # temperature 显式传入时覆盖 constants 默认（一次性 CLI 可指定更确定性的采样）
-        self.temperature = temperature if temperature is not None else constants.DEFAULT_TEMPERATURE
+        # L3：CLI/调用方 temperature > 档案级 default_temperature > 全局默认
+        self.temperature = temperature if temperature is not None else (
+            self.client.default_temperature
+            if getattr(self.client, "default_temperature", None) is not None
+            else constants.DEFAULT_TEMPERATURE)
         self.history_manager = history_manager
         # enable_memory=False 时置空，注册层（_register_tools）依此跳过 MemoryTool，
         # 使记忆工具 schema 不出现在请求中，LLM 无从调用——干净禁用而非运行期拦截
@@ -2902,7 +2909,7 @@ def parse_args():
     parser.add_argument('--no-save-memory', action='store_true', help='一次性对话不注册记忆工具，禁止写入记忆数据库')
     parser.add_argument('--show-reasoning', action='store_true', help='一次性对话在 stderr 显示思考过程流（默认不显示）')
     parser.add_argument('--worker-only', action='store_true', help='一次性对话以 worker 纯执行面模式运行（7 种执行工具，无记忆/人格/派生）')
-    parser.add_argument('--model', type=str, metavar='NAME', help='覆盖配置中的模型：档案名（models 节，如 kimi）或裸模型名（同端点换模型）。服务模式与 --once 均生效（进程级覆盖）')
+    parser.add_argument('--model', type=str, metavar='NAME', help='覆盖配置中的模型：档案名（models 节，如 kimi）或裸模型名（同端点换模型）；歧义时可用 model:NAME 强制裸名 / profile:NAME 强制档案。服务模式与 --once 均生效（进程级覆盖）')
     parser.add_argument('--temperature', type=float, metavar='T', help='覆盖默认采样温度（0.0~2.0，越低越确定）')
 
     return parser.parse_args()