Effective coefficient of restitution should be clamped between 0 and 1

[alice-i-cecile]

Currently, you can set it to e.g. 2.0 (with max combine rule) in the move_marbles example and things will explode.

[alice-i-cecile]

You can observe pathological behavior on the negative end with min combine rule + a value of e.g. -999.

[Jondolf]

Yeah I think this would be a good idea. I can come up with a few potential use cases for bodies gaining kinetic energy on contact, like a game where a ball gets faster each time it hits something, but this might not be considered good practice due to the explosion risk, and I think it's better to do this manually since it allows for more control.

It seems that Unity clamps each material's restitution coefficient between 0 and 1 regardless of the combine rule, so you can't have e.g. 0.5 and 2.0, which would be 1.0 with CoefficientCombine::Multiply. Instead, it will be automatically changed to 0.5 and 1.0, which would be just 0.5.

The main question is, should we clamp both coefficients individually like Unity does, or just clamp the effective, combined coefficient? The second option allows you to multiply e.g. 0.5 and 2.0 to get 1.0, but the first option might be more explicit in that the restitution coefficients are explicitly set to 1.0 even if you try to use a larger value, instead of internally using a separate clamped value for the effective coefficient of each contact.

For explicitness and clarity, I'm leaning on the first option, but I can see why the second option could be useful as well.

[alice-i-cecile]

I think we should do the first for explicitness as well. It also lets you move the checks to construction, which should be much faster.