How, exactly, are you proposing we get the data on "real-world settings" without trying out the approach that's been lab-validated? Barring evidence in the "this might be dangerous" direction, those lab results certainly appear to support recommending people adopt the practice.
Turning this sort of CYA language that's standard in medical research into a scary gotcha is exactly the sort of propaganda the anti-vax folks engage in all the time.
I believe clinical trials are the gold-standard mechanism by which specific questions about interventions may be answered. I’m not pointing out a “scary gotcha.” I’m pointing out an omission identified in the linked work.
Lab confirmation of double masking can be performed in an afternoon. A clinical trial is going to take tens of thousands of self-reporting "I wore two" people for months or longer, plus difficulties with all sorts of confounding variables.
I'm very comfortable with a tentative recommendation based on the lab data for a non-dangerous intervention. If we get observational data a year from now confirming it, even better.
Turning this sort of CYA language that's standard in medical research into a scary gotcha is exactly the sort of propaganda the anti-vax folks engage in all the time.