In the comment above from Eugene Grybinny he states that you won't be able to publish with ignored errors; however, this is only the case if the issue generates a validation warning, which should be documented in the failure for the publish operation.
For the generation of an object that duplicates elements of another objects, but is not a 100% clone, there should not be an issue. I've done this when doing exports to CSV files and then importing those CSV files via -batch mode on mgmt_cli.
If you don't specify the ignore-warnings or ignore-errors CLI option, a -batch operation may complete, but will not execute the changes/additions of the problem items, which can cause follow-on issues doing multiple import operations in succession.
The utilization of the ignore-warnings and ignore-errors option is especially necessary when working with network and group objects, since Get Interfaces in R80 and later will import the static routes and generate a network object for the interface if there is no existing network object covering the target network, and will group them into a set with static route identified networks to add in the topology configuration. If these Check Point generated groups and networks exist, then not using the ignore options will result in failure to create the item from the API, since those networks already exist in another object.