I don't know about all the details of Bell tests using methods like parametric down conversion, but at least in some versions of the EPR paradox you get two photons moving apart in opposite directions. I wonder if you can look for detection coincidences by using photographic plates instead of coincidence counters? The idea would be that if you had two photographic plates on opposite sides of the source, you would get some instances where you would have a perfect matchup of reduced silver crystals, or dots. And that if you inserted crossed polarizers in front of the plates, the coincidences would disappear. I wonder if this analysis is correct, and if so, whether such experiments have been done or proposed?
Yes, something like that is possible, and they've done it in Anton Zeilinger's lab. It's not a Bell's inequality test, and to get it to work it's more complicated than what you're describing, but: they do see spatial (i.e. position) correlations/interference disappear or not based on measurement of one of the entangled particles.
It's described in a nice review paper (which has a bunch of other nice EPR/Bell experiments): Rev. Mod. Phys. 71 S288 (1999)