I have already made that determination on which way sounds better.I note that the OP accomplished the selection in a day or two. In decades of tube amp ownership, I've never found it to be so cut and dried, often taking many weeks / months to come to a determination.
I find myself switching back and forth all the timeWhist I have marked levels on program for decades, I settle on a transformer tap for a particular speaker for a selection of well recorded material and be done with it. One might argue that if one continually switches, neither is ideal.
Examine Fig 1 response curve here in https://www.stereophile.com/content/mcintosh-mc501-monoblock-power-amplifier-measurements-0 for an AutoFormer MC501 and Fig 1 here https://www.stereophile.com/content/audio-research-reference-150-power-amplifier-measurements for a Ref150 and this https://www.stereophile.com/content/parasound-halo-jc-1-monoblock-power-amplifier-measurements for a SS JC-1
Now examine the schematic and impedance/phase diagrams for Stereophile's simulated load here https://www.stereophile.com/reference/60/index.html. [Oh, how I wish Stereophile would run 'transformer' amps into the simulated load from all taps.]
Compare how the three WELL REGARDED amps react to the impedance curve of the simulated load: JC-1 barely, MC501 some and Ref150 plenty. Amps do not have a constant output impedance across the power band. Coupled to a non-constant loudspeaker impedance, the combination of the two are a tone control.
Looking at the phase diagram for the SF Olyiii, we see it leading the 500 to 2k region and abruptly lags @ ≈3k. Given a selection of transformer amps, some may sound best on 4Ω, others on 8Ω [2Ω/16Ω anyone?] and some not at all depending how their impedance reacts with that of the loudspeaker IN THE SYSTEM.
Combined with program phase and level anomalies one has the makings for a Sisyphean task of finding an ideal solution.