I understand the rationale behind this approach, but it falls prey to false equivalence. "Cars" don't kill for the most part, and efforts are taken to reduce the numbers of fatalities caused by road accidents. One only need look at the number of people killed by traffic collisions on the road for this. Seat belts, drivers licences, self-driving technology etc. are all methods implemented (or that could be implemented) to reduce this number. If one proposes that restricting Wahhabist preachers would serve to reduce the likelihood of terror events then it should not immediately be dismissed as a proposal because "more people die of X, so why bother?". In an academic sense, it boils down to a problem of the value of risk minimization.
In the process of minimizing risk, one should not ignore potential black swan events (such as terrorist attacks) as being of orders of magnitude lower probability and thus not worth addressing. Certainly this is not the approach one would take in the process industries, for example, where probability of risk must be conflated with the background population to generate the likelihood of (or expected value of) impact.
I don't understand why terrorism should be any different. Perhaps living in a large city ("part and parcel") makes one more aware of the risk, as now the background population of events is the one directly relevant to my locale. When I hear of terrorist incidents occurring in places that I visit on a regular basis, it certainly makes me more aware of the issue, even if the probability of it occurring to me is very small.
.