My short answer is that I don't know, but I should imagine that the freights of autumn 1998 would indeed have run into the Barton branch proper and reversed out again at Oxmarsh Junction as they were quite short.
Both the up and down lines at Barton Road (named as Barton Lane on site) crossing were made bidirectional when the crossing was rebuilt in December 2015 but not for the whole length of each line.
Apologies for delayed response.
Any freight traffic for New Holland would have run onto the Barton single line and reversed behind Oxmarsh Crossing's signal 27, reverse onto the up main, stopping behind shunt signal 14 to access the freight terminal.
Neither line between Ulceby and Oxmarsh is signalled for bi-directional use. There are procedures to allow for wrong line running in times of perturbation, I would imagine that this would only ever be implemented if a train failed on the up line and needed to be recovered. In theory, single line working could also be implemented for example to work around engineering work on the other line for example, but I think it very unlikely that would happen on the Barton Line. It would certainly not be used for a routing freight working,
The automatic half barrier crossings at Barton Lane and at Butterwood are indeed equipped with the necessary treadles and signage to allow for a wrong direction move to take place, however neither line is bidrectional at this point and single line working, working to a point of obstruction or similar working methods would need to be deployed to allow a train to run in the wrong direction over either crossing. I am sure that there are people more profficient in signalling principles than I who would be able to provide a more succinct explanationn than I have with my limited rudimentary knowledge.