@dantrainman
As far as I understand it, the way padding works (especially with the sleeper), is like this, say we have station A and station B. There are four ways of getting between the two directly, some much longer and abstract than others. The sleeper is booked to take the longest possible time between these two stations, but with no booked calls. Therefore on the day the sleeper can take any four of these routes and as plenty of time has been allocated the sleeper wont lose any time. If it takes the shortest route then it may well have to wait at station B for its allocated departure time, if it has taken the longest route, it may not have to wait long, if at all. This is done so that drivers can keep up to date with routes incase of diversion etc etc.
However with your argument against the WLL, it seems to be a well known fact the sleepers are slightly too quick and a later arrival would be welcomed, so therefore the WLL would seem like a viable alternative to add some extra time on, plus, in the station, if any problems were to occur, no platforms are cocking up SWT operations and all the space below that used to be passport control etc would make a nice waiting lounge, or space for restaurants etc. I also disagree with your logic saying more problems could occur, yes they could, but at that time of night (for the departure at least) the FGW could divert away from the WLL and down via Reading, and wheras thats harder on the GWML out of Paddington. Plus if you consolidate them into one single "super sleeper" franchise, there could be a problem on the WCML and one of the GWML diesels could help out and divert it WLL-GWML-Oxford-Banbury-Chiltern Mail Line-Camp Hill/Kenilworth-Birmingham New Street and onwards.
I think it would be a good idea, speed is not needed with sleepers, I also reckon continuing on the WLL to the NLL then the ECML up to the east side of Scotland, an Inverness/Aberdeen combined. You could even run it via the MML to maximise a longer journey.