feat(vision): 实现 mingxi-vision 推理服务核心功能

- engine/defect_classes: HRIPCB 6类缺陷定义 + 4类扩展 + severity排序
- engine/base: DefectBox / DetectResult 数据类 + BaseInferenceEngine 抽象基类
- engine/ultralytics_adapter: YOLOv8 .pt 推理适配器 + warmup
- engine/annotator: 按 severity 着色的标注图生成 + base64 编码
- engine/loader: 单例管理 + asyncio.Lock 推理串行化 + 热重载支持
- config: Pydantic v1 BaseSettings，读取 .env
- schema: DetectResponse / HealthResponse / ReloadRequest 等响应模型
- api/detect: POST /api/detect，内存读图，推理锁保护
- api/health: GET /api/health
- api/model: POST /api/model/reload 热重载接口
- main: FastAPI lifespan 启动加载模型

测试通过：/api/health 200，/api/detect 空图返回 pass，标注图 base64 正常

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

mingxi-vision/api/detect.py

@@ -0,0 +1,56 @@
+import uuid
+
+import cv2
+import numpy as np
+from fastapi import APIRouter, File, Form, HTTPException, UploadFile
+
+from engine.loader import get_engine, get_inference_lock
+from engine.annotator import annotate, to_base64
+from schema import DefectBoxSchema, DetectResponse
+
+router = APIRouter()
+
+
+@router.post("/detect", response_model=DetectResponse)
+async def detect(
+    image: UploadFile = File(...),
+    conf: float = Form(0.45),
+    return_annotated: bool = Form(False),
+):
+    raw = await image.read()
+    arr = np.frombuffer(raw, dtype=np.uint8)
+    img = cv2.imdecode(arr, cv2.IMREAD_COLOR)
+    if img is None:
+        raise HTTPException(status_code=400, detail="无法解码图像，请检查文件格式")
+
+    engine = get_engine()
+    async with get_inference_lock():
+        result = engine.detect(img, conf=conf)
+
+    annotated_b64 = None
+    if return_annotated:
+        annotated_b64 = to_base64(annotate(img, result))
+
+    return DetectResponse(
+        task_id=str(uuid.uuid4()),
+        verdict=result.verdict,
+        defect_count=result.defect_count,
+        max_severity=result.max_severity,
+        avg_confidence=result.avg_confidence,
+        duration_ms=result.duration_ms,
+        image_width=result.image_width,
+        image_height=result.image_height,
+        model_version=result.model_version,
+        defects=[
+            DefectBoxSchema(
+                class_id=d.class_id,
+                class_name=d.class_name,
+                class_name_zh=d.class_name_zh,
+                confidence=d.confidence,
+                severity=d.severity,
+                box_xyxy=d.box_xyxy,
+            )
+            for d in result.defects
+        ],
+        annotated_image_base64=annotated_b64,
+    )