OldTimer, can I offer an incomplete thought I've been tossing round for a while...
(For the record, you know I'm speaking theoretically and not experientially, as I won't be starting on small cell trials until next season. )
The magic 4.9 number has concerned me for a while. If it is such an effective threshold, then there has to be a 'why' to it.
Particularly in light of the argument sometimes put forward of 'but if my bees can get smaller, why won't the varroa will just adapt and get smaller too".
So I started to think in terms of varroa history, and original host of apis cerana. It's widely reported that varroa does not (I'm not sure how categorical that 'not' is,) infest worker cells of cerana. Next question: size of cerana drone and worker cells. I can't recall offhand the average worker size - somewhere under 4.5 mm - 4.4? 4.3?, but I have found that cerana drone cell size is commonly reported as 5.1 mm. Bingo maybe?
At 5.0 you're under that cell size. 4.9 would give a better margin, obviously, but the difference that would make is going to be hard to quantify.
As for the 'why', maybe the 'why' is just 'because that's the cell size varroa need to breed effectively at their natural size'. If cell size is a key issue and varroa haven't been able to size down to reproduce in cerana worker cells, then maybe they won't be able to size down to reproduce in (very) small mellifera cells.
Obviously there are a bunch of other variables not addressed - duration of capped period, temperature, etc... but it's something to think about.
I do hope you continue with the experiment. It's obvious you have a wealth of experience and an analytical enough head on your shoulders to draw some good conclusions out of it.