Don't laugh. Up here in Brisbane with our lighting I found myself having to ETTL to stop the clouds blowing out when I wanted cloud detail quite often with my E-P3. It happens rarely with the E-M5. The fact is, however, that when faced with a scene with an extremely wide luminance range and some bright areas in which you wish to keep detail, the idea of ETTR can easily become ETTL if you wish to keep those highlight details. Put another way, if we could always manage to do without having to ETTL in order to maintain highlights. then I suspect there's also be a hell of a lot fewer HDR photo composites being made.
------------------------------------------------------------------------------------------------------------------------------------------------
To the original question, yes I've done similar tests a couple of times and find them very handy. I shoot RAW and the camera histogram and highlight blinkies warning do not give an accurate indication of where clipping actually occurs with a RAW file. In practice I can recover highlights that the camera shows as clipped with no trouble at all in Lightroom. I originally did this sort of test for the first time some years ago when using Aperture and found the camera warnings reasonably accurate to slightly optimistic but then repeated the test after swapping to Lightroom and found I could recover more highlight detail in Lightroom and that I therefore had a bit more highlight latitude with Lightroom. I've also done similar tests a couple of times comparing different metering modes in my camera so I know how the warnings work when I swap between the whole screen modes and the spot and spot highlight modes.
I think everyone works differently, and the nature of light can differ in different situations. It can be harsh and contrasty or soft and more diffuse, and that has an effect on highlights. If you're concerned about highlight detail it helps to get a feel for what you can and can't get away with in your sorts of light and with your work practices. I think everyone should occasionally run tests. Not necessarily this test but tests related to where they find issues getting their own images to look right. Experiment with bracketing in problem situations, find out what sort of exposure lets you get the best results. Save your test images. If you change your processing technique or software, take another look at those images and see if your conclusions are still valid or if you can now get a better result some way. Tests don't have to take a lot of time or be difficult to do, but I think you end up getting better results if you occasionally do some tests for yourself.
As for ETTR, I don't subscribe to it as a technique. The reason for that is simple. The ETTR approach is to expose as far to the right as you can without blowing highlights so if you follow that principle sometimes you expose a fair bit to the right, sometimes a little, sometimes you take the "correct" exposure as determined by the meter, and sometimes you can find yourself actually exposing to the left in order not to blow highlights. I prefer simply to expose for the highlights, simply to give them as much exposure as I can without clipping them. Doesn't matter what the range of the scene is, I simply try to get the highlights right and let everything else fall where it will. That lets me capture the widest range of luminance I can and it seems to work for me. That's my approach.
