Release notes need to explain any or all of:

• the change in migration behavior: more frequently, users will see migrations occurring simultaneously with migration cleanups
• the effect on scheduled balancing windows: more migrations may now occur in a given time window, and deletion operations will "leak" past the close of the window for longer periods of time
• possible behavior change in existing user-migrate scripts: moveChunk commands may return sooner than previously
• the fact that more data can be orphaned during a crash of the primary, if multiple delete jobs are queued