Agreed, but I think it is a case of the bean-counters overruling the operators. My suggestion could quite possibly reduce the delays to passengers, but the accountants are only concerned with delays that they have to pay for. Spending money to reduce delays that their company doesn’t have to pay for is not their concern, even if it improves the overall ‘customer experience’. Nor are they interested in retaining passengers beyond the end of their contract.
As regards PNBs: the crew on the train that is on time are still going to have to wait until the other train arrives. They don’t get back to their depot any earlier.
It would cost a fortune though. You're talking about ~200 traincrew needing to learn new track, and once learned they'd most likely need at least an extra 2 days a year off trains to remain competent, to slightly reduce the delays that only occur very rarely anyway.
In the 6 years that that was one of the lines I worked as a guard, I never once encountered a scenario where crossing at Llandovery or Knighton would have improved the punctuality of my train - but to be able to do that, I'd have required a week of training (and given I could only do one round trip a day that's being generous), and 12 days over the course of my career to refresh the route. 12 days when I could otherwise have been used gainfully to work a train.
Multiply that by 200, and add in the extra time drivers need for route learning over guards, and it starts to get kinda pricey.