Implement storage provider fallback for assets

[bedeho]

I am getting a substantial share of videos looking like this

I was told by senior community people that the cause is that some storage providers are having trouble with all the new load, and they are re-synching with greater capacity (which apparently takes many days already!!), and Atlas does not deal properly with this.

Could we instead introduce some behaviour where Atlas tries another storage provider if there is a failure?

[bedeho]

It turns out that there is no retry behavior for any asset, lets introduce that as a backstop for something not working properly in the infra.