132Khz is far beyond what humans can hear, yes. It’s not beyond what modern samplers can do (I’ve seen up to 192kHz). For most applications, it’s insane to go really above 96kHz. Some might say 44kHz is enough and call me out on the Nyquist Frequency Theorem but having a lower noise floor means that you can do other naughty things to audio without raising the noise floor to audible levels. One example is halving the speed of a drum loop which would cut the sampling rate in half (which ends up being 48kHz when sampling at 96kHz of course, which is still above 44kHz so no problem there). For most practical purposes, 96kHz fills those needs.
But when people want to REALLY much with the audio, 192kHz could make some sense, although at that point, your noise floor is really 132kHz when using the above power supply. I can’t imagine needing 192kHz in my applications so I never record at that (I do record and highly endorse recording at 24-bit precision, even for old noisy vintage chips like the SID). So, though it’s wildly pathological, filtering that noise out is in those applications might not be a bad things to look at.
So far, the only graph I have seen is the one I previously. If I am reading that correctly, we know that tantalum can attenuate 10kHz and we could infer from the graph that it can do more since the filtering was approaching zero. That’s a large assumption though to go form 10kHz to 132kHz though 
As far as the SID sound, as I said before, I know it’s noisy, but it’s not a distracting sort of noise and the whole point is to capture the imperfections of the SID - not the PSU
Long story short, for most people (likely including myself), it sounds like this switch-mode PSU is well worth it by itself without having to do any extra filtering. The exception is those with OCD
EDIT: Here is a another PDF describing characteristics of different capacitors when uses to attenuate HF noise from switching supplies. Figure 11 shows filtering over 100kHz (if I am reading it right, the ripple is a bit higher than that). Tantalum does indeed filter out a good chunk of the noise at those frequencies. Ceramics can also work well according to the docs, though they may introduce some noise as well (the document does not say what type of noise or at what frequency).
So, the takeaway is that, again, for pathological cases, adding a tantalum cap seems like the way to go or perhaps a ceramic. Perhaps even the two could be coupled together *shrug*. I don’t even have a way to test the results at those frequencies so, again, it’s all very pathological
