The TenantMigrationDonor's steps should be wrapped in AsyncTry loops to keep retrying until success, the donor node is stepping down or shutting down, or the step generates an error that cannot be retried on and should cause the donor to make an abort decision.

I am not currently aware of any errors that should lead to an abort decision, so for now we can assume every error can be retried on. Perhaps we will uncover some such errors through passthrough testing.

See example of using AsyncTry.