feat: agent proactively re-engages users on returning sessions

Two complementary changes to reduce conversation cold-start friction:

A. Returning-user re-greeting (backend)
- When WS reconnects to a non-empty conversation and last_message_at is older
  than chat_re_greeting_idle_hours (default 6h), the agent emits a warm
  continuation message that references prior history instead of staying silent.
- Self-debouncing: the AI message updates last_message_at, so reconnects
  within the window will not re-trigger.
- Skipped while profile collection is still pending.

D. Topic suggestion chips (backend + Expo)
- New WS message type topic_suggestions carries 3-4 quick-start chips derived
  from the current memoir stage's empty slots (deterministic, no extra LLM
  cost). Sent alongside opening / re-greeting / resume.
- Expo chat screen renders a horizontally-scrollable chip row above the input
  bar; tapping a chip sends the chip's text as a user message and clears the
  row. Sending any text/voice also clears the chips.

api/app/agents/chat/interview_agent.py

@@ -23,6 +23,7 @@ from app.agents.chat.prompt_context import ChatPromptContext
 from app.agents.chat.prompts_conversation import (
     SLOT_NAME_MAP,
     get_opening_prompt,
+    get_re_greeting_prompt,
 )
 from app.agents.chat.reply_limits import (
     nonempty_segments_or_fallback,
@@ -503,3 +504,118 @@ class InterviewAgent:
         except Exception as e:
             logger.error("生成开场白失败: {}", e, exc_info=True)
             return ["你好呀～ 又见面了。今天想从人生里哪一小段回忆开始聊聊？"]
+
+    async def generate_re_greeting_message(
+        self,
+        conversation_id: str,
+        memoir_state: MemoirStateSchema,
+        idle_hours: float,
+        user_profile_context: str = "",
+        background_voice: str = "default",
+        occupation: str = "",
+        profile_birth_year: Optional[int] = None,
+        profile_era_place: str = "",
+    ) -> List[str]:
+        """老对话回访问候：用户带着已有历史回到对话时，AI 主动做承接式开场。
+
+        与 generate_opening_message 的差异：prompt 明确告知有历史 + 距上次的时间感受，
+        要求轻轻引用历史里的具体细节，不能用首次见面式硬开场。
+        """
+        if not self.llm:
+            return ["上次聊到的事我还记着，今天想继续往下讲讲吗？"]
+        try:
+            narrative_state = narrative_coverage_state(memoir_state)
+            control_state = interview_control_state(memoir_state)
+            empty_slots = control_state.prompt_empty_slots_for_stage(
+                narrative_state, memoir_state.current_stage
+            )
+            empty_slots_readable = [SLOT_NAME_MAP.get(s, s) for s in empty_slots]
+            persona = normalize_interview_persona(settings.chat_interview_persona)
+            prompt = get_re_greeting_prompt(
+                current_stage=memoir_state.current_stage,
+                empty_slots_readable=empty_slots_readable,
+                user_profile_context=user_profile_context,
+                persona=persona,
+                background_voice=background_voice,
+                occupation=occupation,
+                profile_birth_year=profile_birth_year,
+                profile_era_place=profile_era_place,
+                idle_hours=idle_hours,
+            )
+            hw = await get_history_with_window(
+                conversation_id,
+                max_pairs=settings.chat_history_max_pairs,
+                max_chars=settings.chat_history_max_chars,
+            )
+            messages: List[Any] = [SystemMessage(content=prompt)]
+            messages.extend(hw.window)
+            messages.append(
+                HumanMessage(
+                    content=(
+                        "（用户回到这个已有历史的对话，还没说话。"
+                        "请基于上文做温和的承接式回访问候。）"
+                    )
+                )
+            )
+            log_agent_payload(
+                logger,
+                "InterviewAgent.re_greeting.prompt",
+                format_history_string(
+                    messages,
+                    omit_system_body=settings.agent_log_omit_system_message_body,
+                ),
+            )
+            re_greet_llm = self.llm.bind(
+                max_tokens=settings.chat_opening_max_tokens,
+                temperature=float(settings.chat_interview_temperature),
+            )
+            llm_t0 = time.perf_counter()
+            with agent_span(
+                logger,
+                "InterviewAgent.re_greeting.llm",
+                conversation_id=conversation_id,
+            ):
+                logger.info(
+                    "event=chat_prompt_built agent=InterviewAgent.generate_re_greeting_message "
+                    "prompt_chars={} history_pairs_total={} history_pairs_windowed={} idle_hours={:.2f}",
+                    _message_contents_char_count(messages),
+                    hw.turn_total,
+                    len(hw.window) // 2,
+                    idle_hours,
+                )
+                response = await re_greet_llm.ainvoke(messages)
+            logger.info(
+                "event=chat_llm_done agent=InterviewAgent.generate_re_greeting_message "
+                "response_latency_ms={:.2f}",
+                (time.perf_counter() - llm_t0) * 1000,
+            )
+            response_text = (
+                response.content if hasattr(response, "content") else str(response)
+            )
+            log_agent_payload(
+                logger, "InterviewAgent.re_greeting.raw_response", response_text
+            )
+            raw_list = segments_from_llm_response(response_text, max_segments=2)
+            if not raw_list:
+                raw_list = [response_text.strip()]
+            max_chars = int(settings.chat_interview_max_chars_per_segment)
+            out = truncate_chat_segments(
+                raw_list,
+                max_segments=2,
+                max_chars_per_segment=max_chars,
+            )
+            log_agent_summary(
+                logger,
+                "InterviewAgent.re_greeting segments={} conversation_id={} idle_hours={:.2f}",
+                len(out),
+                conversation_id,
+                idle_hours,
+            )
+            segments = out if out else [response_text.strip()[:max_chars]]
+            return nonempty_segments_or_fallback(
+                segments,
+                fallback="上次聊到的事我还记着，今天想继续往下讲讲吗？",
+            )
+        except Exception as e:
+            logger.error("生成回访问候失败: {}", e, exc_info=True)
+            return ["上次聊到的事我还记着，今天想继续往下讲讲吗？"]