Braking a helmet by hitting a rollcage is not that unusual (for example here:
https://www.ewrc.cz/images/2009/suma...jf-crash-4.jpg ). If a camera is present on the helmet in such accident it's likely that some of it's parts penetrate into the helmet and you don't want to have metal joint or screws being pushed in your helmet by the rollcage. Moreover hitting a rollcage with the camera is far more probable than without simply because the camera takes some of the space between the helmet and the rollcage.
Massa's helmet was penetrated by 700 grams heavy spring being hit @ 250 km/h which ricocheted, i.e. only part of the energy was spent on the actual penetration. In case of a helmet hitting the rollcage what is moving is the whole head with the helmet and the camera together, i.e. the moving mass has roughly 6,5 kg which means that it doesn't need so high speed for similar effect (6,5 kg heavy head @ 82 km/h has the same energy as 700 grams heavy thing @ 250 km/h). The speed in which the helmet can hit the rollcage is hard to estimate but it can get close to the vehicle speed at the moment of the crash, i.e. it is not a magnitude lower than in Massa's case.
For your information a 200 grams heavy thing @ 200 km/h has a 3/4 energy of a 9 mm bullet and 1/2 more than a 0.32 ACP bullet (Škorpion submachine gun for example). That is not negligible at all like you imply.
Anyway any added weight on the head makes the risk of injury or even death higher. That's a fact hence why it's a bad idea to have a camera on the helmet no matter how well it is fixed. If there is a need for camera on the helmet than it shall be as light as possible, as small as possible and fixed on a helmet which is actually tested together with the camera.