You can't do both. Low leakage gives high capacitance, and vice versa. The underlying explanation is you're shifting around the characteristic impedance of the transformer -- a transformer is a complex LC component with impedance and bandwidth.
What you're really after, then, is a higher bandwidth transformer: both low leakage and capacitance.
To that end, you need less winding length. A thicker core helps, and so the stack height being relatively high is not too bad (and the toroid being almost "square"). It's at least not a bad candidate.
Now, not needing much bandwidth in the first place, is of course a big help. You don't really need terribly low leakage at 50Hz. I wouldn't be surprised if allocating the windings to separate sectors would work out okay. The leakage of this configuration I think will be worse than for a bank-wound (E-I core) transformer. But that might still be okay?
Note that, if you do arrange the windings by sectors, you need the core grounded, so it acts as a shield. Otherwise, obviously it's just a capacitor between the windings, and that's no help. (If you can't get a good connection to the core -- may be the case due to the stacked-rings design* -- you may want to wrap the core with insulation, and then with shielding tape that makes a slitted turn, and ground that.)
*Yeah, that is a bit unusual; I wouldn't expect it to be much more expensive than an EI stack though. They recycle the waste, or maybe they even punched smaller rings out of the holes, who knows.
You can still do a shield with overlaid windings, put a slitted foil layer between them and ground that. That doesn't do anything to reduce the capacitance from each winding to ground (the bandwidth is about the same), but greatly improves the winding-to-winding capacitance, which is a big advantage with respect to mains isolation and noise.
Next best is to put height between the windings, but for a toroid that's pretty annoying. You'd need some kind of spacer, like a layer of thick fiber tape or something?
Tim