It looks like the fuel without oil gives bigger bubbles than the fuel with oil... Can you confirm that?

I know, oil has a stabilizing effect on gasoline (fuel mixed with oil can easily be kept for one year or more, fuel without oil can expire within 3 or 4 months). Which exact oil do you use?
There might be differences with different brands of oil. I have exclusively tested with Castrol PowerRS 2stroke (in some countries known as Castrol TTS), and so far I have never heard of issues with different brands of oil, but I also never have tested if oil brand makes a difference or not.

Another thing that CAN be of effect: How high is your fuel curve, and how far is the main needle open? A wide open main needle and the solenoid doing all the work, seems reasonable to think cavitation will be worse as well, because the pressure drop focusses on the solenoid. A low pressure zone combined with the stirring action of the rattling valve disk, is a rceipe for cavitation. With the needle doing most of the pressure drop, there is less deep low pressure zone on the solenoid, which thoretically SHOULD reduce cavitation.

Now that I think of it, I do remember, in the beginning of developing this system, I did encounter issues with "random wide needle opening" and tuning the engine only on the solenoid, and it worked much better when the main needle is opened no more than strictly necessary.
I normally tune the engine, with unplugged driver but the solenoid installed, to near max RPM on the needle at full throttle (like, the first click that gives a noticable RPM drop from max) and only then connect the solenoid and start acting on the curve. If I understood correct, you took the curves from Guillaume, and maybe because of that, the high speed needle setting is not optimal?
Mind you, this is NOT becaus there is anything wrong with Guillaumes curve, it is because fuel curves are not transferrable from one engine to another.

Exhaust pressure can be tested: With a proper adjusted (for max RPM), full throttle running engine, disconnecting th muffler pressure should result in an engine cutting out due to fuel starvation, and an engine tuned to max RPM without muffler pressure, should start running noticably richer when mufflerpressure is conncted.
I don't think the length of the mufflr pressure connection is an issue, I have planes with longr connections (up to 25 cm). But check if all nipples are open. Sometimes the nippl on the muffler is clogged up with burned oil from the time the engine still ran on methanol with castor oil.
Or maybe the craptrap leaks and the muffler pressure does not reach the fuel tank.
For short duration tests, you can connect muffler directly to the fuel tank. For permanent use, I strongly recommend to always use a craptrap, but of course, if the craptrap leaks, ther will be no pressure in the fuel tank.