Trolley problems appear not just in philosophy - some psychologists are using them in experiments as well. Here is one recent example.

In this case, at least, I think that many of your objections to the trolley problem don't apply. The researchers really are not interested in the ethics of deciding to sacrifice a fat man, they are interested in how the decision to sacrifice might change when the decision maker is on various drugs. And they already have brain imaging results for the trolley problem - so of course they would want to use the same problem in this experiment.