>Should governments ban all activities which surpass a certain risk threshold?
The purpose of regulating personal activities is to mitigate negative externalities to society, not to eliminate personal risk. Thus, whether and how something should be regulated is a function of both how probable it is to cause said externalities, and the overall absolute impact of them.
The negative externalities of skiing or mountain climbing are that others in society have to pay for healthcare costs associated with injuries sustained from these inherently risky activities. However, on an absolute level, skiing/climbing injuries are rare enough that society has decided to eat this minimal cost.
Smoking, on the other extreme, is much more common and has externalities that extend beyond healthcare costs to the smoker [*], e.g. secondhand smoke and litter from cigarette butts. Thus, on an absolute level, its externalities have been deemed unacceptably large by society. But rather than banning smoking outright, there are other ways to mitigate these externalities (bill smokers directly for smoking-related healthcare costs, restrict smoking to areas that minimize secondhand smoke exposure, have draconian fines for littering, etc.)
This argument can also apply to alcohol (e.g. draconian penalties for drunk driving) or unhealthy food (bill healthcare costs associated with obesity directly to the patient) [*].
[*] Yes, I am aware of studies that claim smoking/obesity causes a net decrease of healthcare costs, since most expensive healthcare occurs in old age, which smokers/obese people are less likely to reach. If indeed these studies are true, that should be factored into the actuarial calculus.