I don't understand the treasury concerns with "revenue neutral" fare reform. It implies that the current system is the optimal for revenue yield, which is extremely unlikely.
It seems pretty self evident that a clean sheet system, based on single leg pricing, fixed seat allocations for intercity trains, and using whizz-bang technology such as machine learning, should destroy the current fares structure in terms of income by maximising the value of every seat mile.
There may be a bit of a drop of revenue for a few months as the system teaches itself how much it can charge for trips and prices are deliberately set to stimulate demand, but then when the trains begin getting back to decent load factors, the revenue should be much greater.