Neither the host nor controller sides in our NVMeoF bits support deleting individual I/O queues, so adding the new parameter is cosmetic (But needed if we ever had a compelling use case. I should probably implement support for it on the controller/target side at some point, but would need to be able to test it). However, validating that reserved fields aren't set is useful. I guess I can actually add some tests based on some recent tests I've started writing for nvmft at least, so I'll probably defer merging this until I can add some tests.