Replace temperature with reasoning effort and retry in-stream errors

- Providers get a default reasoning effort (low..max, empty = not sent);
  chat can override it per message. OpenAI sends reasoning_effort,
  Claude sends output_config.effort with adaptive thinking, and thinking
  blocks are replayed during the tool loop.
- Show a thinking indicator while waiting for model output.
- Surface in-stream upstream errors (e.g. overloaded) instead of
  breaking the SSE stream, and retry them up to 3 times when nothing
  has been emitted yet.

backend/src/alphapulse/models/ai.py

@@ -1,6 +1,6 @@
 from datetime import datetime
 
-from sqlalchemy import BigInteger, Boolean, DateTime, Float, ForeignKey, Integer, LargeBinary, String, Text, UniqueConstraint, func
+from sqlalchemy import BigInteger, Boolean, DateTime, ForeignKey, Integer, LargeBinary, String, Text, UniqueConstraint, func
 from sqlalchemy.dialects.postgresql import JSONB
 from sqlalchemy.orm import Mapped, mapped_column
 
@@ -23,7 +23,8 @@ class AiProvider(Base):
     max_tokens: Mapped[int] = mapped_column(Integer, default=8192)
     # 上下文上限（token），超过时拒绝发送而不是静默截断
     context_window: Mapped[int] = mapped_column(Integer, default=200000)
-    temperature: Mapped[float | None] = mapped_column(Float)
+    # 默认思考强度；为空表示不传，由模型决定。Chat 可按轮覆盖
+    reasoning_effort: Mapped[str | None] = mapped_column(String(8))
     is_default: Mapped[bool] = mapped_column(Boolean, default=False)
     created_at: Mapped[datetime] = mapped_column(DateTime(timezone=True), server_default=func.now())
     updated_at: Mapped[datetime] = mapped_column(DateTime(timezone=True), server_default=func.now(), onupdate=func.now())