Oh, the fud of it. The behavior is well documented here https://jira.mongodb.org/browse/SERVER-14766 and in the linked issues. Seasoned users of mongodb know to structure their queries to avoid depending on a cursor if the collection may be concurrently updated by another process. The usual pattern is to re-query the db in cases where your cursor may have gone stale. This tends to be habit due to the 10-minute cursor…
While I agree that it's good to know the limitations of the tools you chose those limitations should be clearly spelled out in the documentation.
I don't think most programmers have the luxury of learning all the voodoo and lore that surrounds MongoDB from JIRA tickets and blog posts.