Object Detection Birds undocumented input data format

From [object_detection_birds.ipynb](https://github.com/awslabs/amazon-sagemaker-examples/blob/master/introduction_to_amazon_algorithms/object_detection_birds/object_detection_birds.ipynb):
```
# Define the bounding boxes in the format required by SageMaker's built in Object Detection algorithm.
# the xmin/ymin/xmax/ymax parameters are specified as ratios to the total image pixel size
full_df['header_cols'] = 2 # one col for the number of header cols, one for the label width
full_df['label_width'] = 5 # number of cols for each label: class, xmin, ymin, xmax, ymax
full_df['xmin'] = full_df['x_abs'] / full_df['width']
full_df['xmax'] = (full_df['x_abs'] + full_df['bbox_width']) / full_df['width']
full_df['ymin'] = full_df['y_abs'] / full_df['height']
full_df['ymax'] = (full_df['y_abs'] + full_df['bbox_height']) / full_df['height']
```

According to object detection algo [documentaiton](https://docs.aws.amazon.com/sagemaker/latest/dg/object-detection.html), the algorithm requires annotations as `"left", "top", "width", "height"`. I couldn't find anything about `header_cols`.

What is the right way to describe bounding boxes? As in this notebook or as in the documentaiton?

Start with introduction_to_amazon_algorithms/object_detection_birds/object_detection_birds.ipynb and compare its bounding-box construction with the linked SageMaker object detection documentation. Clarify the expected bounding-box format and the purpose of header_cols, then ensure the notebook and documentation describe the same input data requirements.

Written by the indexing model from the issue text.