I think the actual answer is to fix the clients that are producing those nevent IDs without encoding the author or relay metadata of where it was published or found. I feel like it’s ok to say we don’t have enough information to find this event on this decentralized network because you didn’t provide it. That’s just my opinion though.
I should have given more info above. This is in the context of an external web app working with a NIP-21 URI in the format `nostr:nevent1xxxx`. Relay and author metadata are optional so we can't always do the Outbox model bootstrapping sequence magic as we may not even have the author pubkey (although I'll tag nostr:nprofile1qqswuyd9ml6qcxd92h6pleptfrcqucvvjy39vg4wx7mv9wm8kakyujgpypmhxue69uhkx6r0wf6hxtndd94k2erfd3nk2u3wvdhk6w35xs6z7qgwwaehxw309ahx7uewd3hkctcpypmhxue69uhkummnw3ezuetfde6kuer6wasku7nfvuh8xurpvdjj7a0nq40 here just in case as he often has magical solutions to everything nostr :))
I wonder what software like njump.me do to retrieve a note by nevent.
Discussion
Agreed, although "fix" may be a strong word given that NIP-21 language makes it optional. https://github.com/nostr-protocol/nips/blob/master/19.md . Still, you are right, clients should include relay metadata when generating nevents