I think the reason this wasn't supported before was because it wasn't supported on all the detection types. Right now it's only really in [`Crop Hints`](https://github.com/GoogleCloudPlatform/google-cloud-python/blob/7ba4d7a211e19d2a3fe8641ce3443e30a34ae4da/vision/google/cloud/vision/image.py#L164). If someone can confirm that this is supported on everything else, I can start adding it.