To solve the wrong example of MultiScaleRoIAlign() (pytorch#1085)

In this [docs](https://pytorch.org/tutorials/intermediate/torchvision_tutorial.html),
the `2 - Modifying the model to add a different backbone` part has wrong example code.

```python
roi_pooler = torchvision.ops.MultiScaleRoIAlign(featmap_names=[0],
                                                output_size=7,
                                                sampling_ratio=2)
```

should be

```
roi_pooler = torchvision.ops.MultiScaleRoIAlign(featmap_names=['0'],
                                                output_size=7,
                                                sampling_ratio=2)
```

reference: 
1. https://github.com/pytorch/vision/blob/master/torchvision/models/detection/faster_rcnn.py
2. https://discuss.pytorch.org/t/index-error-using-custom-backbone-on-fasterrcnn/79834/4

Co-authored-by: holly1238 <[email protected]>

Author: seriousran

intermediate_source/torchvision_tutorial.rst

@@ -263,7 +263,7 @@ way of doing it:
    # be [0]. More generally, the backbone should return an
    # OrderedDict[Tensor], and in featmap_names you can choose which
    # feature maps to use.
-   roi_pooler = torchvision.ops.MultiScaleRoIAlign(featmap_names=[0],
+   roi_pooler = torchvision.ops.MultiScaleRoIAlign(featmap_names=['0'],
                                                    output_size=7,
                                                    sampling_ratio=2)