feat: retry failed blob transfers due to timeouts once during imports
During gitlab#350920 (closed), we're observing that migrations for large repositories are failing due to random timeouts when transferring blobs (sample correlation ID: 01G4C1HWYZ5YNJA3BRGWDM8TKR
). To avoid wasting all that time and end up with a failed migration (retried three times to no avail due to timeouts when transferring different blobs) we should retry the transfer once in the presence of a context deadline exceeded or network timeout.
Edited by João Pereira