It may be Apple´s fault, not RME's... It is probably caused by big.little CPU architecture and OS wrongly allocating certain threads to  efficiency cores instead of power ones.....

Similar problems may happen on latest intel CPUs on Win. But on win user may switch off efficiency cores or finetune what program uses what cores....

It seems to me it is not possible on Apple...

No, if that is the case, you should open a complain ticket with Apple and ask them to fix it!
If you search around the web, it does not happen only with RME interfaces. It happens with other manufacturers, as well.
And there are many people that don't have any problem, at all.

I get it in the daw or when using the computer normally.

At the moment I turned the high security back on and installed the latest V4.09 again.
High security was off all these time. Which is also needed for V3 usage.

It is going better. Some hiccups once in a while so not completely solved.

I have tried this and unfortunately does not fix my stuttering on my M1 Max / UFX III

Mixing with the UFX III is absolutely a pain.
I came across a text regarding a so called bug in Logic and delay compensation calculation. That it only counts the pdc at the start of the session or when you reset the audio driver. I find this hard to believe although I do find it possible with the RME.

If I add big latency plugins I get immense stuttering when I hit play, this is solved when I reset the audio driver in Logic's audio settings but reappears when adding plugins.

The thing is, I can mix without any errors when using the build in headphones. So something is messing up buffers with the RME.
In sonoma there's an orange microphone at the top right showing. With the stuttering is flashes.

Is there any solution, explanation or even an investigation from RME?
Do I need to sell my UFX III because this is not workable?

I think your portrayal is technically not correct (looks as if you put a statement out of context), see longer text below *)

In terms of audio loss. More likely is what we read the last months in this forum about all the issues. Partially it were Apple problems, but some problems seemed to be caused on kind of "upgrade installations" of users, where a complete new installation could cure some or all problems.

*) You see it e.g. clear with Cubase under Windows. Decrease ASIO buffersize, and you will see an increase in (EDIT:) CPU and ASIO load.  What happens with smaller buffer sizes is simply a higher interrupt load, which is always bad for a system.
The principles are all the same, regardless of which CPU, which OS.

Because the interrupts block CPU cores for other activities and the CPU also has to switch the context much more often and reload all CPU registers with the other context. That's a lot of work for a CPU (management overhead).
Imagine you had to empty a bathtub at a predetermined time interval (e.g. 5 minutes) with a bucket or a glass of water. I think with a water glass you would work up a sweat to get the same amount of water filled in the same amount of time.
All channels are always transmitted on the computer/recording interface. The data must be collected in a timely manner. If you put additional stress on the CPU with small buffers during a high workload, then dropouts are more likely to occur.
The art is
1. to use good drivers that do not occupy a CPU for too long, because the drivers have the highest priority and the process scheduler has no control over the length of time the driver stays on the CPU core (due to data integrity, Windows and MacOS are not real-time systems
2. taming of background processes, to lower the system load, to have more headroom for audio processing in time.
3. to adapt the buffer sizes to the needs of the work/project. When recording, use high buffer sizes (so that nothing is lost). Middle range when mixing/mastering. When playing via VSTs, smaller buffers so that the RTL stays under 10ms.

Tried class compliant mode. Still happens with the RME, tried different usb-c ports and even a hub. My guess it needs a firmware update.

A €298 Motu M4, same usb-c port, no problems at all.