It's not really heavier cars as such that are safer, more larger cars. Making the car heavier means you've got more energy to dissipate in a crash, making it larger (longer) allows the energy to be dissipated over a longer period of time. The main safety 'benefit' of heavier cars is that they cause more damage to what they hit and thus come to a stop more slowly, be that another lighter car or a deformable object. Fine if you're in the heavier car, less good if you're not. I would be very interested to know how occupant safety varies with size and weight for otherwise similar cars in a two-vehicle crash. I suspect that a head-on crash between two heavy vehicles would greatly reduce the safety advantage they have relative to average vehicles.
I'm not sure how good an example F1 cars are - their safety comes from having a very strong tub and impact-absorbing barriers around the circuits. The drivers are also very tightly constrained compared to road cars. Two F1 cars having a head-on crash at 30mph could be more violent than two road cars having the same crash as there's not a lot of energy absorbing material between the drivers. F1 cars don't tend to run head on into other cars or brick walls though...
Making the passenger cell out of high strength, light weight (and very expensive) materials would help, as reducing the car's overall mass means there's less energy to be dissipated in the crumple zones. In turn this should mean the crumple zones themselves can be lighter (but not shorter as there's no getting away from the need to minimise deceleration rates), further reducing the mass of the car. Alternatively they could be engineered to withstand a greater impact, but would this come at the cost of harsher low-speed impacts?
Anyhow, I think it should be possible to make a light car that can withstand single-vehicle or lab-test crashes as well as a larger car, the problem comes in the real world when you start crashing into heavier vehicles.
Re: aerodynamics - if the car is intended for city use, then aerodynamics aren't especially significant. That doesn't mean aerodynamics aren't important for a vehicle with 40bhp - there are bonneville streamliners that do 180mph with that amount of power, aero drag is still the limiting factor. What I don't know is how much engineering effort is involved in getting a car design down to a Cd of ~0.3 in the first place. My gut feeling says it might be harder than you expect, in which case the wind tunnel testing would be justified. Anyone? For any car, shaving 0.01 off a Cd of around 0.3 is going to make a fairly negligible difference, there are greater gains to be had by reducing the overall Cd.A figure by reducing the frontal area.
FWIW the reynolds numbers of car and person sized objects are such that aero drag increases with the square of speed for all 'interesting' speeds, but at low speeds constant friction and rolling resistance losses are more significant.
"Rockets are just another name for trouble. Either you just had trouble, you are having trouble, or you are going to have trouble." - Milt Rosen, Viking Program Director, White Sands Missile Test Range