I appreciate the research, but I'm not sure what could possible be done to change the situation materially. "Making more accurate contamination maps" strikes me as only useful in a really best case scenario.
A very fair point, but in this case installing filters or a reasonable hope for decontamination isn't really an option; the only thing we can do is prevent, detect, and evacuate.
Which makes faster detection a huge deal - I'd imagine one of the conclusions drawn from the results will be how best to detect ASAP without putting sensors literally everywhere.
I'm not sure that's feasible. It sounds like releasing an agent anywhere in the system is going to lead to it being pistoned through the entire system. You don't even need to do it at a stop. You'd need sensors at stops though, and presumably in the tunnels all along the way to sense a release. Nothing else would give more than a second or two of warning, that would initiate a lethal panic.