You need to stop thinking about 'relief' lines. Current operating practice between Oxford and Wolvercote is that container trains use the central pair of tracks as much as possible and are held there while passenger services run past them on the outer tracks.
You keep going on about 'huge expense'. I repeat, we are not talking about starting with a blank canvas. The land needed is already within the Network Rail boundary all the way to Radley (and beyond on the west side of the line to the former Abingdon Junction). The basic track formation remains in place, though obviously in need of some tlc here and there but reinstating the fourth track from Oxford to Wolvercote was a fairly straightforward process, so why should it be any different heading in the other direction?
The capacity issues between the Didcot and Oxford areas go back a long way. A grade-separated junction at Didcot East would be nice but very expensive indeed, while providing more track for as much of the distance to Oxford as possible is a much more straightforward exercise. Just sticking a loop in somewhere along the way, with the inevitable speed/time penalty to trains while getting in and out of it, could actually make matters worse.
Keeping a container train rolling while passenger trains get past it, with it then following in their wake, is a rather more practical way to operate. If quadruple track extended to Radley (which would admittedly require an expensive require an extra bit of bridge over the Thames between Oxford station and Hinksey) then you get an eight-mile stretch to do that in, rather than the three miles from the station's south end to Wolvercote.
