Performance issues related to $nin/$ne querying in large database

Question

I am working on a pipeline where multiple microservices(workers) modify and add attributes to documents. Some of them have to make sure the document was already processed by