Yes we have studied actual rates of infection in response to interventions like air filtering, so these studies account for all that real-world complexity and messiness you worry about.

The article is complaining that every study doesn't redo the evidence collection, end-to-end, every time. That's not realistic and not necessary.

A lot of your specific questions are leading (with a nothing-we-can-do attitude underneath) or asking the wrong question (eg expecting one universal number for "hours of filtration per infection prevented").

For instance the correct answer may be air filters in classrooms and buses and workplaces, but strangely your line of questioning doesn't even consider that possibility.

This would be like someone in the 1800s questioning how handwashing avoids Cholera if they don't wash their hands at home. I think I see a solution to this one...