It may seem that I am trying to avoid your actual question, but that is not the case, really. First, most transformer-coupled tube amplifiers provide a choice of 4, 8, or 16 ohm taps, not 2, 4 and 8 ohms. Second, the very low impedance at very high frequencies exhibited by an ESL is not so much of a problem for the amplifier, because very little energy is required at those high frequencies. (For example, I made impedance vs frequency curves for my Sound Lab ESLs after eliminating the crossover components; impedance does not drop off significantly until 5kHz and goes to 2 ohms only above 10kHz. But every ESL will be different from every other one, in this respect.) I suggest that you try each pair of output taps for yourself and see what you prefer. There is no right or wrong answer to your question, except to say that most amplifier energy will be required in the bass and midrange frequencies.
Nordicnorm, M-L candidly admits that they design their speakers to favor SS amplifiers. (See also my post of 11-02-16.) This means that there is circuitry at the input that dramatically lowers the impedance seen by the amplifier, well below the native impedance of the speaker itself. Thus your finding is not surprising. This is why I sold my last pair of M-L ESLs about 20 years ago and never looked back. No way I would give up my OTL tube monoblocks just to suit one particular speaker.