If there's any significant amount that you distribute throughout the timetable you run a very big risk of it eliminating white space and the possibility of running other services, or they'd have to run late, etc etc etc. At the end is the safest place for big lumps as there's very little chance of any kind of knock on like that, but what performance time there is tends to be quite minimal, an Edinburgh - KX (and vice versa) will typically only have a total of 3 minutes, and that's usually distributed in 1 minute allowances in three totally different places anyway.
Which rather misses the point, or rather confirms that the point is to tick a box and avoid a fine/DR payment rather than to ensure robust connections.
If we look at Avanti from Glasgow to Euston, say, the connections at intermediates are what need to be robust - some of them can be quite short and infrequent. Arrival at Euston doesn't matter as much as most people are finishing there or connecting to very frequent buses or Tubes, while people who are continuing across London have itineraries with very loose connections indeed.
Thus if you were designing for journey robustness, you would pad intermediately on that route.
By constant a LNR to Euston from Northampton needs a bit at Bletchley and Watford for connections but not really at other intermediates because there aren't really any connections, most people will be walking, cycling or getting in their car.
All it needs is a bit of common sense and a willingness to do things for passenger benefit rather than ease of operations, something the railway often fails at.
)