feat: 完善模型管理功能

- 新增模型 API 路由，支持 CRUD 和测试连接
- 支持 MiniMax、GLM、OpenAI Compatible 三种供应商
- 添加连接状态持久化 (untested/connected/disconnected)
- 修复 CORS 和数据库模型兼容性问题
- 前端 UI 优化：供应商默认 API 地址自动填充

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

backend/app/schemas/model.py

@@ -0,0 +1,41 @@
+"""
+Model Schema
+"""
+from pydantic import BaseModel, Field, ConfigDict
+from typing import Optional
+from datetime import datetime
+from uuid import UUID
+
+
+class ModelBase(BaseModel):
+    """Base model schema"""
+    provider: str = Field(..., description="Model provider: minimax, glm, openai")
+    model_name: str = Field(..., description="Model name")
+    api_key: Optional[str] = Field(None, description="API key")
+    api_base: Optional[str] = Field(None, description="API base URL")
+    is_default: str = Field(default="false", description="Is default model: true/false")
+
+
+class ModelCreate(ModelBase):
+    """Model creation schema"""
+    pass
+
+
+class ModelUpdate(BaseModel):
+    """Model update schema"""
+    provider: Optional[str] = None
+    model_name: Optional[str] = None
+    api_key: Optional[str] = None
+    api_base: Optional[str] = None
+    is_default: Optional[str] = None
+
+
+class ModelResponse(ModelBase):
+    """Model response schema"""
+    id: UUID
+    created_at: Optional[datetime] = None
+    updated_at: Optional[datetime] = None
+    project_id: Optional[UUID] = None
+    connection_status: Optional[str] = Field(default="untested")
+
+    model_config = ConfigDict(from_attributes=True)