Piaggio P-119

It might be that one of the issues with the Piaggio's engine cooling problems was derived from the inlet design. Modern jet inlets usually have a separation between the fuselage and the inlet. The idea is to not be swallowing any disturbed and low velocity boundary layer air and only be tasing in free stream ram flow.

Look at the drawing in Post #38 again.
In order to get equal cooling you need the same volume of air at the same pressure hitting the faces of the cylinders all around the engine.
With a scoop offset to one "side" (downside/bottom) of the engine to air going to the upper cylinders has alonger, more twisted path than the air going to the lower cylinders.
a splitter on the intake is not going to change that.

Now figure that the scoop and internal passages have to maintain that even airflow at warm up on the ground, full throttle climb at much less than full speed, full speed and part throttle at high speed cruise.

That is an awful lot to ask from an offset inlet. Please note that had the engine been placed a few feet further back (impossible due to center of gravity) you have a lot more room to straighten the airflow out.

I would note that many radial engines in much more conventional installations had one or more cylinders than ran hotter than the others.

In a good cowling design the front of the cowl acts as a higher pressure plenum and the aft of the cowl will act as a low pressure area with some evenness to channel the air flow due to differential pressure equally between the cylinders. The ideal not often achieved! For instance in the R985 installation in the DHC2 Beaver cylinders 1 and 9 (the ones at the top) run hot, this is particularly noticeable with a slight change in AOA. My guess for this (WAG) is that the cowl exit in front of the cockpit windshield is considerably narrowed. The R2800 also had a couple of perennial step children cylinders, though I don't off hand remember which ones.