Make list_blobs resilient to random failures (e.g. retry policy)

[jorgecarleitao]

I am observing that when listing blobs, sometimes the listing fails in one of the pages, resulting in having to restart. For a large number of blobs, I am being unable to list all blobs, as I need to restart the listing from the start.

An error on the Paginated results should return the continuation token, and we should be able to create a new pagination from the last token (or something equivalent).