I've actually seen the exact same thing on one of my own P7100s, with latest firmware (J2R18-something and F9-something DSP) programmed in trunking mode but with no affiliation and all talkgroups TX inhibited.
It happens so rarely that I have not been able to gather additional data on the problem.
If I caught this problem 'in the act' then I will try to use the site statistics features of the FCC menu to determine if, maybe, the problem is related to ProVoice activity occuring on a specific repeater. And then watch that repeater for additional traffic and read the frequency error out for that repeater.
I'm going to guess that one or more repeaters is far enough off frequency relative to the receiving radio's own reference frequency that it may cause decoding problems.
But that would not indicate a system problem, necessarily. It may be something as simple as the repeater being within spec but a bit off frequency, and the radio also being in spec (or close) but it may be off frequency in the other direction.
I would also take a VERY close look at the logical channel list(s) and be on the lookout for a channel that is incorrectly entered in the list. The last time I had an issue with missed calls, I found that one of the channels in the logical channel list was entered as 8xx.1250 when the correct value was 8xx.1125.
I will note that I BELIEVE that the radio will be more sensitive to a frequency error in ProVoice mode than in clear mode. Thus, the problem, if it IS a small frequency error, might not appear on an analog transmission while still causing problems in ProVoice.
So it'd miss every single time that logical channel number was called up. Simple, logical, easy to fix once you know what to do.
I didn't quite trust the published frequency assignments for the local sites so I went to the FCC ULS database, did a geosearch at the tower site, and got the whole list of site frequencies. That positively identifed the frequency error.
I could also set up a spectrum analyzer near the tower site and capture every operating frequency as they come up, but that's tedious work when you're trying to identify the actual transmit frequencies of 20 repeaters.