The idea is that a student who is reading from beginning to end will always be provided with a train of thought. The student will have no motivation to stop, and memorize an unexplained product of thought.
In many cases, it may be excessively ambitious to hope for students of multivariable calculus to learn to avoid making mistakes that were made by the original creators/discoverers, so it would make sense -- if some parts of the actual historical path of discovery/invention are used -- to include consideration of ideas that don't work, although it may not be immediately obvious that they don't work. Trains of thought don't proceed by the shortest path from beginning to end, but will naturally involve some dead-ends and changes of direction, but the changes of direction occur only after somebody detects a potential error, and confirms that there is an error.
It's true that students who have learned to succeed by memorizing may fail to read carefully, and may memorize some errors, but perhaps there is more to be learned from errors than from unexplained products of thought that were created by means of mysterious, undisclosed trains of thought.