There is no overhead in the post recalculations regarding related questions as there is no ad hoc structure to hold them.
In fact, they are just dynamically calculated when they are displayed in the widget. The logic uses a score assigned to each question and the score determines how much a question matches a given question. Then the questions with the higher score are displayed.
If you are interested in how the score is assigned you can take a look at the qa_db_related_qs_selectspec function. You'll see the score involves matching words in question titles, tags and categories.
So I wouldn't advise you to go deep in there to decrease post reindexing times.