This ticket is being filed speculatively in the event that we end up not returning a cursor from bulkWrite. If we keep the cursor, then I believe we can just remove the TODO and close this ticket.

In the getWriteSizeFn passed to targetWriteOps for non-bulkWrite batched writes, we ensure that for unordered writes we do not build any batches that could possibly generate an error response too large to fit into a single reply.
I think we don't need this for now in the equivalent bulkWrite logic due to the cursor response allowing us to return a response of arbitrarily large size, but we could end up needing this if we change plans, so I wanted to file this ticket so we remember that.