Hmm, that's an interesting discovery, or as most folks would say "a real bummer".
You're right on the oil feed system and frankly I can't see how oil would end up where you're finding it under normal conditions. The feed is the front (diagram below), it goes into the cams and then floods the head. If there was a leak at the block/head interface then I'd expect to see oil outside the block, not coming out of #3 & #4 exhausts.
Although there's a drain from the head to the sump and a vent to atmosphere, both of which you've shown are in place, there's still a reasonable pressure in cam covers. If you're curious, take the filler cap off when it's running and stand back..... (no, don't try this at home, it will get messy)
So the cam followers are always flooded and even with vents in place there's enough pressure to find any weak points - look how often you see TCs with cam cover gaskets/cover nuts misted with oil.
As you're not burning oil then my first thought would be exhaust valve guides. As you say there's no seals and it's all down to machining the exact tolerance required so in theory one or two could have been reamed incorrectly. But that's a long shot, if all the guides have been replaced there's no logic that says one or two would have been machined differently to the others.
The other option, even more obscure, would be an incorrectly machined valve stem, again I can't offer that with anything other than the 1:1,000,000 chance of being right !
So if it's machined correctly then I'd look at the cam followers and sleeves.
They don't usually wear because they run in oil and on the early heads in fact the followers didn't have the sleeves the later ones had, but the oil is getting into #3 & #4 exhausts somehow.
I have read somewhere about keeping the followers in the correct order when you get the cams off and making sure they go back into their respective sleeves, so maybe there's a wear pattern involved. I honestly don't know and the reason might be the cam lobe/follower interface and nothing to do with the follower/sleeve interface, but I'm throwing that one in as a freebie.....
Personally I'd run it a bit longer and give it a few more coats of "thinking about" before pulling the head, ever hopeful that it will go away. Yes, I'm an unrealistic optimist at times.
If the oil is annoyingly persistant then I think my first move would be to look at the cam followers, lift the cams and measure the followers and sleeves on all the valves to see if some are better than others. If the followers are slack in the sleeves on #3/#4 then I could see more oil getting to the backs of the guides than on #1/#2, and I'll bet there's enough pressure from the cam cover and pumping action of the follower to push it down the guide into the exhaust port. You wouldn't see oil burning because the valve is only open when gas is flowing out
If all sleeves are identical spec then it's head off and finding somewhere who can pressure test it on the bench. I think I'd ask QED myself, they have a long history of TC heads so they might have seen similar and perhaps even a weld repair ?
Fingers crossed that swapping cam followers around fixes the problem.....
Brian