A Go Driver user has found that download a large file using GridFS results in a CursorNotFound error if their application requires more than 10 minutes to process the first 16MB of the file. The GridFS spec only allows setting the batchSize and noCursorTimeout options for the Generic Find on Files Collection API, not the downloadToStream/openDownloadStream APIs. Can we mitigate this issue by setting noCursorTimeout for the find on the chunks collection? The cursor would still time out after logicalSessionTimeoutMinutes, but it's better than 10 minutes and retains the "stream" abstraction without exposing cursor-specific details in the API.