The majority of road cars have had the screen bonded in for well over 30 years.Nothing to do with airbags [though it possibly helps]
They do do a lot for stiffness though. In the 70s full size Holden screens were vulcanised in and when they came loose [regularly] you could feel the car flex. Rebond the screen and it felt tighter and did not squeak either!
These days most windows are glued in with Uerathene and seldom come loose. Though on occasion still leak and when they do often rust in that area.

In the mid-seventies I used to sell silicone sealant for Dow Corning. Urethane adhesives were just making an appearance in the consumer market.

One sidelight: I wondered why they didn't use silicones for windshield wipers. The problem was silicone fluids bleeding out of the rubber and smearing the windshield. I just saw some advertized so I guess they have the problem fixed after 35 years!

I don't know actual numbers, but I do know it's a big deal in GT Porsche's to have the proper windscreen installed and bonded. You can't just throw in a piece of Lexan and call it good. It's a big deal in terms of car performance. That's why you see any reasonable team carrying about a 1/2 dozen of the things around. It's a pain, but absolutely mandatory.

Identifying which bit of the body is flexing is most easily done in CAE. In practice static testing is much more difficult than dynamic testing, so what we usually do is correlate the CAE to the dynamic test (ie frequencies and mode shapes and modal masses) and then run the static test in CAE world.

Mazda like to hang strings or wires across their cars and go for a drive in them. They do learn a lot from this, whether they'd learn more from a rig test I don't know. They have a subjective attribute called 'structural feel' which is related to this.

We don't test unglazed BIW, so I can't provide numbers. I'd guess it would be a 30% hit at least.