The references to USB sockets and most types of SMPSUs associated with networking etc seems to be a red herring, as they must provide electrical separation between the AC input and DC output. Therefore, they are no more likely to inject a DC current into the circuit conductors that would saturate and disable a type AC RCD, than any other electronic load. On an SELV supply, there would have to be an input-output fault (nasty, possibly live iPhone!) as well as a fault to earth from the other side of the DC supply, and the unit would still have to carry on functioning to produce the offending DC. Possible, but in practice the whole thing would likely blow to smithereens and no longer generate DC. The input rectifier of an SMPSU can produce a DC load if one side of the non-separated DC bus leaks to earth and one leg of the bridge blows out (or it's half-wave), but this should not result in shock because the electrical separation remains, so it doesn't rely on the RCD for protection. It could disable the RCD from protecting against other faults, but again the 99% likely result is that the bridge will simply short out completely and blow the internal fuse / circuit OCPD, leaving the RCD to function normally.
Phase-angle controlled loads are more likely to have a DC component in their supply current and there were, maybe still are, appliances that deliberately produce one. For example, hairdryers with two heat settings that use a diode to halve the element mean current, draw their entire load as DC on low heat. Does this prevent operation of a traditional type AC RCD? I would think it probably does, without there even being a fault. I would actually like to try this (anyone care to do it?) Put a half-wave low heat thingy on a type AC and run a normal RCD test?
I'm familiar with RCD selection on the basis of DC and HF load current components from theatrical and machine installations. Large lighting dimmers can produce a significant DC component in their supply current, especially in the event of an output device fault leading to half-wave loads of many amps. For these, a type A may provide adequate protection, but for a device that can fault from the DC buss of a non-separated bridge rectified mains supply, e.g. an inverter, the smooth-DC sensitivity of a type B is needed.
In summary, yes, food for thought if you're not familiar with DC-sensitivity of RCDs, but I am not fully in agreement with some of the comments.
