If the mempools stay consistent across nodes, then any bloom filter based lookup would still work, but orphan rates might be worse than if we have graphene **with** CTOR.
The orphan rates without better compression (which is achieved with CTOR and Graphene) WILL be worse.
