It's known to be harsher with plasmas. For a time hdtvtest.co.uk did both the camera an LB tests to compare, check their reviews database.
It is not "harsher" on plasmas - plasmas have inherently higher latency than LCDs, CRTs, or OLEDs.
Plasmas update the image globally - they buffer a frame then update the entire panel at once. (or at least in <1ms)
The minimum latency possible at the top position would be 16.67ms at 60Hz for a Plasma display with absolutely zero processing latency.
Of course like all digital displays, there is going to be some amount of latency on top of that.
If a Plasma TV had 23.33ms internal processing delays, the meter would read 40ms at the top and 40ms at bottom of the screen.
CRTs, LCDs, and OLEDs scan the image from top-to-bottom, which means they can start drawing the image line-by-line as it is received, rather than waiting to buffer an entire frame before the panel can be refreshed.
With a CRT that has effectively zero latency, at 60Hz they should measure ~0ms latency at the very first line on-screen, and ~16.67ms for the very last line on-screen. (1000/60 = 16.67ms)
With an LCD or OLED display, you will typically have the same 16.67ms difference between the top and bottom of the screen at 60Hz, but there are probably additional processing delays on top of that, affecting the top and bottom measurements equally.
So if they had a 23.33ms processing delay it would measure 23.33ms at the top of the screen and 40ms at the bottom.
However it's also possible to buffer the frame and do a fast scan-out with an LCD or OLED. (and apparently some LCD/OLED displays can refresh globally too)
So you would have that 16.67ms minimum latency like a Plasma TV because you have to buffer the frame, and then you might scan out the image in say 8.33ms total rather than 16.67ms.
For an LCD that had zero latency but did an accelerated scan-out of 8.33ms, it would measure 16.67ms at the top and the bottom would be 25ms.
If that display had 23.33ms processing delays, it would measure 40ms at the top like the plasma, but 48.33ms at the bottom.
Average latency measurements are meaningless, and the middle reading with the Bodnar tester doesn't really tell us anything useful about the properties of the display.
The top measurement on its own tells us approximately how much processing delay the display has.
The top and bottom measurements together can give us useful information about how the panel is updated and what the maximum latency should be.
OLEDs have a lot of potential for gaming - they could be the closest thing to a CRT as far as latency and motion blur is concerned, but I have no hope of that happening when LG are the only company mass-producing TV-sized panels.