But, it's hard to know how the power resistor is constructed so there may in fact be something in the construction of that type of resistor that has a brick wall limit of current that's in the single digits...
Or maybe there was a manufacturing flaw in that resistor. Don't know. It was one of those gold colored aluminum/ceramic types so there was no way to get into it. I didn't even think to try, I just realized that I'd put too much current through it and moved on.
Meaningful brick wall current limits are rarely found in resistance elements. However brick wall energy limits are common.
Consider a wirewound or foil resistance element of some known resistance alloy. The ultra short non-repetitive pulse power limit is directly related the thermal mass of the resistance element itself, and for a known alloy can be estimated from the energy that would be required to raise the resistance element to its melting point.
For longer pulses the thermal mass of the former or substrate becomes significant and once you get out to the whole second timescale, its the thermal mass of the whole resistor.
However stresses in the element may cause it to neck near its melting point causing hot-spotting on subsequent pulses, it may gradually oxidise, and thermal degradation of its insulation may produce byproducts that corrode it, so the repetitive limit will be considerably less.
*SOME* resistor manufacturers have good datasheets with pulse power ratings, typically 5 to 10 times the continuous rating for whole second timescale pulses, and a few even have application notes: e.g.
http://www.ttelectronicsresistors.com/pdf/application_notes/Pulse-Overload_AN.pdfA mains rated heating element will certainly withstand the pulse power of a mains soft-start application, but it would be advisable to add a thermal fuse or cutout rated for an inductive load in case the shorting relay fails.
For bench use, you can also use NTC thermistors designed for anti-surge applications, but for reliability reasons, even though they may be rated for continuous operation, one should use a shorting relay, which also reduces the risk they will still be hot, and thus ineffective if the supply is briefly interrupted. One *ASS*U*ME*s the user will be smart enough not to repeatedly frobulate the upstream switch, as that can cause worst-case core saturation with the thermistor(s) already hot enough to be near minimum resistance.
Controlling the surge while staying within the resistor's pulse power limits is $EXPENSIVE$ and either requires a very high wattage resistor or a higher value resistor (to limit the peak power) than would initially be expected. The purpose of a bench isolating transformer does not and should not include soft-starting its load, as the load should have its own soft start circuit if required. As such, the use of a second relay to delay load switch-on till after the anti-surge resistor has been shorted and the transformer current has stabilised, is likely to be cost-effective at higher VA ratings, as a much smaller resistor can be used if it only has to handle the unloaded transformer surge.