From the video previously posted by "highnfar," here are screen caps of images taken with the backlight on during metering, and backlight off during metering....
Yes, I agree I can see a visible difference between the images. But which image is the correct exposure for the scene and what you are trying to achieve? Without a review of the histograms and images, neither may be correct, and for night photography nor should they be assumed to be correct. (FWIW, the image on the right looks over-exposed, but its almost impossible to tell from such a small grab - and I do know the left is the LCD affected exposure.)
I'm not saying there isn't a problem, there clearly is but it looks to be manageable. In my experimentation, I've found that the only time the exposure is affected is when the LCD panel in is brighter light than the image to be captured and how much it is affected depends on the intensity difference.
With this knowledge in hand we are forewarned and hence forearmed and can work around it - albeit we shouldn't and I wish we didn't have to, but it is what it is at the moment.
My workflow for night photography will not change:
1. Compose the image and set focus.
2. Set a high ISO, Aperture Priority, biggest aperture for the lens, set autobracket +/-1 stop.
3. Review the 3 histograms and images.
4. Switch to Manual mode, with same ISO, same aperture, and adjusted time based on review of previous images and the result I want to achieve.
5. Confirm adjusted exposure by creating another image or adjust again. Determine Bulb exposure time for desired ISO and desired aperture.
6. Set bulb setting, start exposure, grab a coffee, plug in the iPod and wait.
The good news is I've confirmed (on my 5D3 at least) that the light leakage does not get to the image sensor when the mirror is up, so the activation of the LCD or the use of a head-torch or similar during the exposure doesn't affect the image.