Deal more elegantly with failures

There are several types of failures (e.g., connection timeouts) which are ignored right now, these should be handled appropriately. E.g., just skipping that asset and logging it so it can be synced later, or depending on the error perhaps adding self-imposed wait times.