feat: retry pre import manifests due to transient errors

Checking the errors defined in #702 (closed), I realized that the timeout is always caught at the top level of a pre-import manifest:

1 error occurred:\n\t* pre importing manifests: pre importing manifest: pre importing manifest: * different errors after this*

So it occurred to me that we can move the retry operation to this top level instead, that way we can retry the full operation. What I don't know is if we would retry to transfer full blobs for layers if they already exist or if we simply move on to the next? If we do the full operation, then perhaps this is not the best approach.

Related to #702 (closed)