The backup/restore process should be more resilient on intermittent errors

[DDvO]

Since #685 got closed without handling all points mentioned, I repeat one of the points here:

On any intermittent (e.g., network) error, backup fails (is not resilient to such issues),
and on retry it partially takes, in total, even more space and time.

When using remote (online) storage, both backup and restore should not immediately fail on data transmission errors,
but only after a reasonable (ideally, configurable) number of unsuccessful retires with suitable timeout periods.