Phil - certainly less noise would indicate less impact on the seat at slow speed. Spring behaviour has a huge effect at high speed regardless of lash in controlling valve bounce. I have seen high speed photography from the Advanced Engine Design Forums presented by Comp Cams, and it is a real eye-opener.
A couple of thousands either way is no big deal. There is certainly more slop in the rocker arm to shaft, and especially the rocker tip to valve stem contact point. I routinely square up valve stems, and profile the rocker tips (sometimes finishing with a whet stone) to produce accurate readings. An engine with any mileage will have a divot on the rocker which defeats extreme accuracy.
The B21 and B230 engines even with a short valve train have a different clearance for hot and cold, as the variable is more the head material than anything else. The pushrod Volvos have an expansion rate between the block and pushrod which seems to be very similar.
I agree that to get any kind of accuracy the engine must make at least two full revolutions. My overlap method requires that too.
I have seen every kind of failure there is in B20-18 valve trains. I can't say why they do it for sure. Cam lobes that aren't there, shattered and mushroomed lifters, worn out pushrod tips, broken rockers, and valves so far recessed that the head was below the combustion chamber surface. Modern oils have improved the lifter/lobe problems, but the valve recession seems to happen "just because". Unleaded fuels are not the problem, that much I have determined.