I stepped away from this for a while because I was getting frustrated, and that seems to have carried through enough to frustrate some of the devs. But I think there's something important this analysis is missing, and that's that none of the support*, looking into the issue, or concluding they couldn't reproduce it, happened in what they designate as their support channel. It instead happened on Twitter (which I don't use), after Patrick posted there, and to a lesser degree on my GitHub issue after I'd posted my frustration here and on their forum.
I wouldn't have been happy, as such, if they'd replied on the issue or on the forum with that information, but I recognize the rest of what you said--it's a F/OSS project, I'm a non-paying user, and it's a relatively low-priority issue (though I don't think I'd entirely agree with that wrt the OpenVPN problem). But that isn't what happened. There were two weeks of silence on the forum, and at least a week on the GitHub issue. I don't think I'm being impatient to have expected something in that time.
They suggest hardware issues, or that the hardware was inadequate. I wouldn't rule it out as a possibility, but I'd note that it's been more than adequate to run pfSense (which also runs ntpd as a service) for 4+ years. And even when I replaced the system with one roughly twice as performant (a Protectli FW4B), the same problems persisted. That might not completely rule out a hardware failure as the source of the problem, but it makes it exponentially less likely.
The last straw was today. It notified me that an update to 20.7.2 was available, so I installed it. No errors were reported, but it left the system (and my Internet connection) in an unusable state. Enough. I put the old SSD with pfSense back in the old box, plugged it back in, and turned it on--I'm back to a solid connection. I think I'm done with OPNsense, at least for the time being.
* I did get some help on my thread about ntpd, but then just... nothing. No "we're looking into it," no "that's strange, it shouldn't be doing that," no "I can't duplicate this", nothing. Just silence for 2+ weeks, until I posted my admittedly-inflammatory (though I still believe mostly correct) summary of their Twitter response.