Note, impedance curves are basically meaningless because they are defined by part geometry. Think of it this way, the index of refraction is approx. sqrt(mu_r). So a wavelength of say 10m becomes some 10s cm in ferrite, and the path length of the core matters.
So you get all sorts of shapes, long/tall tubes, short rings, etc., with corresponding different peaks and humps in the impedance curve.
It's actually hard to select/design a core for this; ring cores aren't usually rated for impedance, so it's hard to know where to start.
Anyway, the HF cutoff is dominated by turn-to-turn capacitance, so the impedance doesn't simply rise forever as turns goes up. Rather, the impedance band shifts lower (and somewhat narrower) with turns. By how much, depends on winding geometry and core materials, though I don't know how much of each, offhand. (EMI CMCs are illustrative, but mostly available in the highest mu / lowest Fc material, so it's hard to gauge how much effect mu/Fc is on overall total.)
It's further hard to tell, because the size of core needed for HF work, may not even fully be utilized, i.e. the losses are so strong, field doesn't even get into the middle; at some point it's worth using stacks of short (washer-like) cores, or tubes, instead of thicker parts. Skin effect does not go away, using ferrite, it's just pushed to higher frequencies. IIRC, ballpark figure is power ferrite, mu ~ 2000, resistivity and hysteresis whatever, 100kHz, ballpark 10cm. So, 10MHz should be around 1cm, about the thickness of say a T240.
But that's mainly a contraindication of MnZn ferrites (~1 Ωm), so just don't use #77 or #31. NiZn (~1 MΩm) are fine up to higher frequencies. Hysteresis loss isn't that much lower so that's still something to keep in mind (loss is loss, hysteresis still works to cause skin effect), but clearly #43 and #61 do well for middle and high frequencies.
I would probably use #43 here, and do some measurements (preferably under power) to determine its impedance. On top of all the grounding and common mode stuff I mentioned earlier, of course.
Although Type-43 material has lower loss, Type-73 gives the greatest impedance over the HF range.
And, mind, given the above, it's not to say #73 (a MnZn material) is worthless; not saying the above (Amidon) plot is wrong or anything.
You can see the above-described effect in the data:
https://fair-rite.com/73-material-data-sheet/notice the double-breakpoint response, mu' drops off shallowly around 1MHz, maybe due to hysteresis, then again at 10MHz due to resistivity. Notice the core size, 18/10/6mm toroid. One or both of these breakpoints will depend on core size and aspect ratio, so we can expect lower breakpoint(s) for practical power cores (35mm+ say).
Over the 1-10MHz range, the impedance will go something like |Z| ~ sqrt(f), which is useful for ferrite beads where you need losses. That's kind of irrelevant here -- not to say it's good or bad, just that it's beside what we're optimizing for here, which is |Z| over whatever desired HF bands.
You see a similar response for #43, but the breakpoint at ~2MHz is steeper this time, and the 2nd breakpoint is much higher (50MHz+?).
It may well be that, in thicker sections, #73 retains more |Z| than #43 at this frequency; I suspect it'll be close at least, and I wouldn't feel bad using either in this range I think.
Tim