If you were to run just one channel of each Ref 75 it would be prudent, IMO, to put load resistors on the output of the unused channel.
Tube amps having output transformers should of course not be operated unloaded when they are processing a signal. But in this scenario, where one channel is not processing a signal, consider what would happen if a tube developed a short or some other problem that would cause that channel to output a "signal" while unloaded, and with no means for the user to become aware that anything is amiss.
I agree with the comments about bridging. Paralleling the channels is another possibility, but I certainly wouldn't do that without an ok from ARC, and I can envision ways in which a cloud of smoke might result. For example, if a small signal tube in one of the paralleled channels were to fail, the other channel would be applying a large signal to the output of that channel, while that channel would be trying to force its output to zero volts. Not a good situation.