docs: fix the formatting json issue of troubleshooting page (#463)

Signed-off-by: Jim Wang <[email protected]>

docs_src/troubleshooting.md

@@ -4,13 +4,12 @@ Q: Why is the performance sometimes on CPU better than on GPU, when running pipe
 
 A: The performance of pipeline benchmarking strongly depends on the models.  Specifically for `yolov5s` object detection, it is recommended to use the model precision FP32 when it is running on device `GPU`.  If supported, then you can change the model precision by going to the folder `configs/opencv-ovms/models/2022` from the root of project folder and editing the `base_path` for that particular model in the `config_template.json` file.  For example, you can change the the base_path of `FP16` to `FP32` assuming the precision `FP32` of the model yolov5s is available:  
 
-        ```json
-            ...
-            "config": {
-            "name": "yolov5s",
-            "base_path": "/models/yolov5s/FP32-INT8",
-            "layout": "NHWC:NCHW",
-            ...
-            }
-
-        ```
+```json
+    ...
+    "config": {
+    "name": "yolov5s",
+    "base_path": "/models/yolov5s/FP32-INT8",
+    "layout": "NHWC:NCHW",
+    ...
+    }
+```