When running notebooks/build_person_directory.ipynb, it will fail after the step of `### Adding and associating a new face`. This is because that the code will need to be updated with real data. This is not ideal as an example. By comparison, C# version of the sample has been improved with sample data: https://github.com/Azure-Samples/azure-ai-content-understanding-dotnet/pull/5 We should fix the Python sample by following the .NET improvement too.