move faiss installtion to prepare.sh

test_tipc/docs/test_serving_infer_python.md

@@ -60,9 +60,6 @@ Linux GPU/CPU PYTHON 服务化部署测试的主程序为`test_serving_infer_pyt
   python3.7 -m pip install paddle-serving-app==0.9.0 -i https://pypi.tuna.tsinghua.edu.cn/simple
   python3.7 -m pip install install paddle-serving-server-gpu==0.9.0.post101 -i https://pypi.tuna.tsinghua.edu.cn/simple
 
-  # 测试PP-ShiTu识别模型时需安装faiss包
-  python3.7-m pip install faiss-cpu==1.7.1post2 -i https://pypi.tuna.tsinghua.edu.cn/simple
-
   # 下载模型与数据
   bash test_tipc/prepare.sh test_tipc/configs/ResNet50/ResNet50_linux_gpu_normal_normal_serving_python_linux_gpu_cpu.txt serving_infer
   ```

test_tipc/prepare.sh

@@ -209,6 +209,7 @@ if [[ ${MODE} = "serving_infer" ]]; then
     # prepare serving env
     python_name=$(func_parser_value "${lines[2]}")
     if [[ ${model_name} =~ "ShiTu" ]]; then
+        ${python_name} -m pip install faiss-cpu==1.7.1post2 -i https://pypi.tuna.tsinghua.edu.cn/simple
         cls_inference_model_url=$(func_parser_value "${lines[3]}")
         cls_tar_name=$(func_get_url_file_name "${cls_inference_model_url}")
         det_inference_model_url=$(func_parser_value "${lines[4]}")