In the future when everything has returned to normal for a shake up of the Sleeper services
I'd split the Lowland sleeper with the Glasgow Central and Fort William portion running as one train and then the Fort William portion would detach and run via the Rutherglen chord.
The Inverness, Aberdeen and Edinburgh portion running as one train, with the necessary portions still splitting, I realise that Edinburgh passengers would have an earlier arrival, but surely provision could be made to use one of the lesser used platforms
As others have said, the usage of the Fort William portion varies significantly. Baring in mind that it also serves as the only regular direct service between Edinburgh Waverley and Fort William, the last thing you want to do is to remove this - as during the quiet periods in the berths, there is still a fair usage of the seated coach between the two. Thus splitting at Carstairs as part of the Glasgow sleeper removes this somewhat important connection, and would essentially kill the seated coach for this service regardless if you were to continue it to London or not.
It would be interesting to see the figures showing how Fort William to England seated demand compares to that of the seated demand for Scottish based journeys.
When it comes to the question of additional calls in general, where the timetable allows, I do see merrit in maximising the calling points for the sake of trying to max out the usage - particularly for the Aberdeen portion with the likes of Laurencekirk, Broughty Ferry, Cupar, Ladybank, and Markinch. Not that it would make much of a difference, but providing these towns with their first direct London service could be an incentive to get people using it without significantly altering the timetable.
I do agree with others' suggestions for more WCML stops such as Milton Keynes. As much as there could be benefit in calling as Birmingham and Manchester, it's not rocket science to know that any further usage wouldn't even make a dent in improving the economics of the sleeper in the first place, and in the long term, running through more congested areas at peak time could further hit reliability. However it would be interesting to see how a direct link to the two in a hypothetical scenario would fare.