Just because the tank is light, doesn't mean that a completed fuel carriage will be much lighter than any other car.
Better gravimetric energy density doesn't make so much difference for a train, so its other advantages -- production on-site, zero emissions, carbon-neutral -- are more important.
At the trainyard, the LH2 would be banked in an underground tank where insulation is easy. It is not necessary to refrigerate it after it is liquified. It will gradually boil off, so you just make more anytime the level goes down, whether from boil-off or pumping out to a locomotive tank.
My expectation was that native electrical energy will be cheap enough that wasting some is better than complicating your system. Each tradeoff will be an engineering analysis. In this case it will all depend on how good your insulation is, which is of course better in an underground tank. Many old conventions will be overturned as the marginal cost of electrical power at peak times gets cheaper.
Both will be tried, so we will see.