Merge pull request #1049 from wangjiawei04/develop

add readme for examples

python/examples/README.md

+## Examples
+
+### Support `--use_trt`
+
+the following models support `--use_trt`, which means you can use TensorRT to accelerate inference at Cuda 10.1 or higher.
+
+- imagenet ResNet50/ResNet101
+- detection faster_rcnn/yolov3/pp-yolo/ttf-net
+
+
+

python/examples/README_CN.md

+## Serving模型示例
+
+### 支持TensorRT的模型列表 `--use_trt`
+
+以下模型支持TensorRT，可以开启 `--use_trt`来加速在线预测，其他模型不能开启。
+
+- imagenet ResNet50/ResNet101
+- detection faster_rcnn/yolov3/pp-yolo/ttf-net
+i