Refactor assistant configuration and database seeding

- Update Makefile to include new database seed commands for assistants and credentials.
- Refactor assistant model to use explicit fields instead of a config dictionary, improving data integrity and clarity.
- Implement new seeding SQL script for assistants, ensuring dependencies on credentials are respected.
- Modify backend routes and frontend components to accommodate the new assistant structure, including direct field access for prompt, API URL, and keys.
- Enhance the AssistantPage component to handle the new data structure and streamline the save process for different assistant types.

backend/db/models.py

@@ -95,8 +95,13 @@ class Assistant(Base):
         String(40), ForeignKey("knowledge_bases.id", ondelete="RESTRICT"), nullable=True
     )
 
-    # 类型专属字段(形态各异):prompt / graph / dify|fastgpt|opencode 端点+key(打码)
-    config: Mapped[dict] = mapped_column(JSON, default=dict)
+    # ---- 瘦类型专属字段(真列,稀疏:按 type 用其中几列) ----
+    prompt: Mapped[str] = mapped_column(String(8192), default="")    # prompt / opencode
+    api_url: Mapped[str] = mapped_column(String(512), default="")    # dify / fastgpt / opencode
+    api_key: Mapped[str] = mapped_column(String(512), default="")    # dify / fastgpt / opencode(打码/哨兵,同凭证)
+    app_id: Mapped[str] = mapped_column(String(128), default="")     # fastgpt
+    # workflow 专属:图(nodes/edges)。要版本化时再迁出到 assistant_workflow 表
+    graph: Mapped[dict] = mapped_column(JSON, default=dict)
 
     created_at: Mapped[datetime] = mapped_column(DateTime(timezone=True), server_default=func.now())
     updated_at: Mapped[datetime] = mapped_column(