There might be some question whether cars less efficient than planes, but I am not aware of any such doubt w.r.t. trains, which is what we are discussing here. Also note that few people would drive the long distances that a plane or train enable them to cover.
> but it is unclear how the method of calculation (ecopassenger.org) works, whether it considers that rail and road infrastructure requires far more energy
The methodology is available [0]. It considers direct emissions, so no maintenance or manufacture. Can you make an argument why these should be more than a low percentage of the direct emission? I mean sure, it might require 1000x more energy to maintain, but its not like that is a large part of the final emission per kilometer.
> since you cannot change the load of plane and train separately
Why would you need to? Of course people who take the train instead of the plane reduce demand for planes and increase demand for trains. So there are fewer plane trips and more rail trips. Assuming that rail trips are environmentally advantages, why would that be a bad idea?
> even though it is clear that flight operators often overbook planes, while trains often run with pretty low loads.
The model considers rail load factors based on real world data where available, substituting 35% as an approximation where no data is available. Do you have any reason to object to that?
0: http://ecopassenger.hafas.de/hafas-res/download/Ecopassenger...