Make the notebooks more reliable in test runs

[ojarjur]

We've recently added the .test.sh script which runs each of these notebooks to validate that none of the cells raise errors when executed.

Conceptually, this is fine, but most of these notebooks were written with the idea that they would be run once rather than repeatedly.

As such, we should go through each notebook and see if there are things we should change to make them more reliable when run multiple times.

As a concrete example, we have at least one issue in the 'tutorials/storage/Storage APIs.ipynb' notebook where it creates a sample bucket, adds an item to it, deletes just that one item, and then deletes the notebook. If that process fails after creating the sample item but before deleting it, then every subsequent run will find that the sample bucket is not empty, and attempts to delete the sample bucket will fail.

That particular notebook needs to be updated to make sure that it better handles the scenario of the sample bucket already existing and there are probably also similar issues in the other notebooks.