Another comment.
Perhaps I'm confused by the use of the term "leakage" wrt AC mains. An ideal power transformer would only transmit 50/60 Hz!
50/60 Hz on output is usually called ripple and the PSRR is the relevant value.
The way I've seen and used leakage has to do with a circuit formed across the power supply via parasitic capacitance (non infinite impedance) and whose return path is the common ground.
These loops can involve multiple power supplies linked by non infinite impedance connections (cables). The reason that SMPS are more prone than LPS is precisely because the higher frequencies have less impedance.
Unfortunately I don't have time for a thorough debate... I'll comment on my perception which clearly is not the same as yours. Not worth a debate to me as my context and career is not this sort of thing.
An ideal transformer would transmit any AC signal of any frequency above DC. If it only transmitted 50/60 Hz that would be a filter, or a transformer with a filter, or perhaps the closest to your definition would be a resonant transformer circuit like the old SOLA units. That does not match what I learned, maybe different schools, or I'm just too old and out of date. Take your pick. But my son's recent college class used the same definition that I use for an ideal transformer.
Ripple can be 60 Hz for a linear supply in the USA, but is also used to define noise ripple for a SMPS. At least in the data sheets and spec I have. Again maybe just a difference in work location or whatever.
PSRR is defined differently in different datasheets, something that has always vexed me. Commonly it refers to DC or very LF AC rejection. OTOH some systems I designed chips for required PSRR be spec'd broadband, well into the RF/mW/mmW region. Ditto CMRR. I always look to see how it is tested in an audio product's datasheet.
To me, in this context (e.g. commercial product power supplies and audio), leakage usually implies DC or LF AC signals, not broadband coupling, though the term "leakage" could certainly imply wideband signal coupling. Signal leakage is a term used in RF mixers to describe LO leakage into the RF and IF ports, for example. Note there are technically capacitive, inductive, and resistive/conductance leakage paths.
I am not sure what you mean in your last statement, sorry. I have not observed SMPS to be more sensitive to leakage, rather the opposite, but they generally generate higher frequency harmonics in higher amplitude than linear supplies. The impedance of traces, decoupling capacitors, vias etc. goes up with frequency, typically leading to higher voltage noise, is that what you are saying?
A little dim today - Don