My argument is that putting forward a hypothetical situation with perfect foresight, ignoring secondary effects, ignoring human nature, and constraining decisions to two options leads to bad thinking.

Yes, that is what you argue for. But an argument doesn't only contain obeservations, it needs a setup where you put the argument in context and a conclusion where you show how your observations relate to your setup. Your setup is that trolley problems are a theoretical tool but your observations all come from real life situations that simply doesn't match and that diminishes the quality of your argument.

On the contrary - I don't think trolley problems are good at testing moral intuitions in theory.

And that is what you have in your setup but then don't substantiate. That is what doesn't follow from your observations.