> there doesn't appear to be any actual clipping.
?? Yes, there certainly is, in the left channel (e.g. around 00:17 to 00:25 and around 01:15 among other places). It's not super-hard clipping, though--there is some slight variation among the peak levels--and the unclipped peaks don't look as if they would have reached more than ~3 dB farther than what you actually see, so it makes sense that you wouldn't hear it. It's also noticeably asymmetrical, i.e. in the positive-going direction of the waveform there is considerably less clipping than there is in the negative-going direction, and that's a mitigating factor as well.
Clipping on simple, steady tones becomes distinctly audible at the 1-to-3% THD level, while moderate amounts of clipping on transient signals (whose overall character is more noise-like than tone-like) is far harder to hear even if it occurs repeatedly. This is why a good limiter doesn't sound "positively bad" if it's used in small doses on appropriate material--it's not so much what you hear as what you don't hear when it works, and what you don't hear, you don't miss.
Again, another reason not to set your recording levels so that the highest peaks are only -12 or even -15 as some people evidently do around here ... it's OK to live a little.
--best regards