It appeared to me that the reason for
the local overheating must be somewhere in the coolant flow. I think,
someone on the other e-mail group pointed to that possibility.
What it boils down to is the fact that
not all the cooling pockets in the rotor housing are open to coolant flow.
In particular, the second opening on the inlet side is totally blocked off
by the side housings. I left it open, and worsening the situation, there is
a comparatively large hole in the rotor housing web. As a result, most of
the coolant did go through this opening, and not through the area around the
plugs. If I had looked close at the flow pattern from MAZDA this problem
could have been avoided.
Fortunately, there is no permanent
damage to the engine, and the engine in the airplane has not run yet. I
fixed the problem by closing off the opening. The engine is now going back
to the test stand for another 10+ hrs of beating.