...many if not most closed wounds do not conceal significant strength loss from decay.
As always, the smaller wound the better, pruning or injecting.
Why is that?
Here's my theory.
Statistically, before decay gets to the point where it causes trees to structurally fail, more than 2/3s of the wood needs to be decayed. One small injury is unlikely to cause that much decay.
Large injuries take a long time to cover over the wound. If it is quite large, it may not close in the life of the tree, and even be the cause of catastrophic failure.
Your observations that wounds that cover, don't cause failure, is not because they cover and decay stops, its because they are small and the area of CODIT is small.
There is not evidence, that I know of ,that would prove that if you had two equal wounds, on a tree, and one had wall 4 close over faster, or perhaps wall 4 held open somehow, that there would be statistically less decay in the wound that 'sealed'.
In fact, there is evidence of just the opposite. Consider Shigo, and his work with artificial pruning sealers. No statistical improvement.
Three things come to mind.
One, is the studies Shigo did with flush cuts. Remember how fast they closed over, but how much worse the decay was?
The second thing is, I wonder if there have been any studies on sugar maple taps? They make a small trunk injury and then keep it open for years, do you think there is a larger area of decay that can be attributed to the tap keeping wall 4 from closing?
Finally, do you recall the study where they wrapped wounds with Saran Wrap immediately after wounding to hold ethylene released by the tree close to the wound area, and got statistically better closure?
I wonder if the trees in this study could be dissected to see if there was any difference from the control, now years later.