Just to confirm that @Emily_Nicholson is right here. This can happen if you delete the last 6.x index and then immediately shut down a node to start the upgrade, because index deletes are a little asynchronous so the node may not have received the message about the last bad index being deleted before it's shut down.
Temporarily reverting the node back to 7.x and allowing it to rejoin the cluster will refresh the stale metadata. Downgrades don't work in general but in this specific situation it's ok because the 8.x node won't have changed anything on disk by this point.
Unfortunately the 8.x node doesn't really have a way to provide this information - the problem is that the bad indices are too old for an 8.x node to even read their metadata which would be required to name them. However I'm not sure it really would help to know those names - you can't do anything different with this information.
That said, I see room for improvement in the error message and opened this PR: