Technical photographs must mimic physical reality. Art may do so if that's what you are after. But they don't have to do so.
I think it is an extremely interesting idea, and I don't even own a TV.
Having, in a former life, worked as a TV repair technician, and having personally administered the violent release of CRT vacuum on numerous occasions, I can testify that most CRTs have a front glass thickness ~1/2" - 3/4" thick, depending on the size of the CRT.
Hence my attempt at doing this resulted in my using a pinhole camera in front of the CRT, where the pinhole size was about the same as the phosphor pixels on the CRT. Of course, a glass camera lens also would work, perhaps with a faster exposure time, but you may have issues with focus at the edge, since most glass CRTs are curved, if you used a wide angle lens and a close-up distance to the TV. You probably would have better results with a longer lens further away from the screen, getting a flatter field.
On old scanning electron microscopes, images used to be recorded by exposing sheet film laid on the CRT display. It used to be acceptable to label images as $largenumberX where X was based on a 4x5 negative's proportions. Now that everything is digital, the "10,000X" doesn't mean anything and you have to put a ruler somewhere in the image for scale.
Hi! this might be a little forward for my first post, but i haven't gotten round to introducing myself yet- i will shortly, i promise.
There's relatively simple way around the glass thickness problem, but then again it might defeat your purpose:
Use/find a plasma or TFT screen.
With a TFT screen hooked to a monitor, you can do "powerpoint shutter" instead of DVD shutter, the top layer of the TFT is really thin compared to the CRT glass, you can do endles color balance, exposure, contrast and so on to the image on photoshop and powerpoint...
But then you lose the cool factor of doing it from a "vintage" TV...