Also need to check if two analyses with the same `Tokenizer` object are occurring concurrently and raise an error in that case.