I think it can still be a potential problem at just 10uA+ leakage if the source voltage is large enough.
Mains leakage (or any other ac noise) can get rectified, it's then much
more of a potential problem because it can charge capacitors (rising until charge current = any load or capacitor leakage current ) up to possibly high peak voltages .
It only takes a couple of diodes which happen to be oriented to form a voltage doubler
( looking for an example of something that one might actually connect to a bench supply would be a bridge followed by a capacitor then into a fixed voltage regulator the type that has a power saving disable pin so it only uses a few uA in shutdown followed by your precious circuit which is earthed say by a scope probe).
That capacitor will now potentially charge upto twice the peak voltage (2 * 1.41 * Vrms) of the ac leakage when the regulator is disabled. In this example If we are lucky the cap will start leaking and limit the peak voltage before the reg gives up the ghost and goes s/c with the resultant surge taking out whatever is after it.
In the example here bridge diodes are used to rectify but transistor pn junctions and MOSFET body diodes may also happen to be oriented to form a voltage doubler (even internally inside ic's when you earth one pin via a scope probe)) so this may not be as improbable as one might think, I haven't tested this hypothesis of possible internal rectification damage to ic's so i got no idea how probable it is.
As the previous posters mentioned for bench supplies that use 50-60Hz transformers the best solution is to use transformer with an inter-winding shield or split bobbin transformer, the shield can reduce primary to secondary inter-winding capacitance (Cint) from a typical 100pF to < 1pF .
If shielded tranny not used then there should at least be a capacitor fitted from the secondary side circuit to earth (47n-100nF 200V+ which is usually put near the supply output) , this forms a capacitive voltage divider with Cint reducing the leakage Vpeak to a safe couple of volts .
I can recall a thread about the rigol dp832 where a.c leakage voltage concern was raised and I mentioned the absence of those earthing caps as problematic, I don't know if they fixed it yet.(
https://www.eevblog.com/forum/testgear/rigol-dp832-firmware-updates-and-bug-list/msg640067/#msg640067)
If you have any supply with excessive voltage a.c leakage then try the bridge rect followed by say a ~1uF cap (rated 100v+) and earthed one side experiment, a minute or so depending on a.c leakage current should see the cap charge to 2*Vpeak (though when connecting a meter the measured Voltage will quickly fall from the Vpeak as cap is partially discharged and the source is now loaded through the 10meg input resistance of the meter) . I ought to mention that disabling supply output with a switch which usually only breaks the +out would not disable the charging of the cap by the a.c leakage current (which is 'common mode', that is it's equally present on both + and - output's) which can still continue to charge via the -Vout socket.
Regards