I will explain why PAllen's curt dismissal of the issue I brought up is unjustifed (and why this is so revealing about the way understanding of the new physics developed).
If Einstein had come up with a thought experiment that showed that something he believed to be true broke the law of conservation of energy, far from considering conservation of energy to be "irrelevant" he would surely have considered this very significant. He would have had to conclude that if the thought experiment was valid either his belief was incorrect or the law of conservation of energy was genuinely broken, which could be as radical a discovery as special relativity.
In March of Einstein's "miracle year" of 1905 Einstein had published a paper showing that light was composed of particles, despite exhibiting wave properties. Obviously, since quantum mechanics was still undiscovered, this was something only partially understood. My question is, was his improved (but still partial) understanding of the nature of light enough in June of the same year (when he published his first paper on special relativity) to have given him the chance to conclude that the energy in a light signal transmitted between two points at different time dilations would change according to the time dilation?
My considered view is he probably could have if he had combined his two pieces of work. He knew that the packets of light had energy equal to h nu. With this frequency clearly being measured relative to a clock, the energy shift is clear, as long as one rejects the unacceptable idea that the number of photons changes depending on which frame one counts them in.
Pretty straightforward with a century of hindsight, but not at all in the year when special relativity and the quantisation of light were born. Given that, two years later in 1907, Einstein was the first person to realize that gravitational time dilation occurred, perhaps he can be forgiven for having not inferred this with a simple thought experiment in 1905!