Well, I was able to filter out the clinking noises from the 1983 mix with a 2khz lowpass filter, and the resulting muffling of the audio didn’t sound different enough from the ROTP to be unambiguously wrong. A 6khz filter wasn’t enough to filter it out, though. Considering the frequency response of 16mm optical probably tapers off before a 6khz cutoff, it’s all plausibly still a limitation of the medium IMO.
To test this, I think I’ll run the ROTP mix through a 2khz highpass. If it contains audible non-hiss detail above 2khz, then I’m assuming we should have been able to hear the clinking sounds. If it’s all hiss with maybe a few barely audible bits here and there, then this was likely a false alarm.