The issue with overload against frequency is the RIAA equalisation that's applied on the cutting side and the opposite on the replay side. This theoretically makes the overall result constant velocity, and as RIAA accuracy can be easily within 1dB and typically rather better than that, measuring the overload at 1kHz gives a decent enough measure of what's important.
That is HIGHLY dependent on the RIAA topology in the preamp. One example- consider a preamp with two gain blocks separated by a passive RIAA filter. The LF will likely not overload the first stage, but the HF very well could. Likewise, the LF could possibly overload the second stage, but the HF is unlikely to. In the real world, I've seen major differences in overload margins as a function of frequency, and that's one of things that's guided my designs.