Wrong generalized inverse masses in solve_vel

[felixbjorkeson]

I am pretty sure that the generalized inverse masses calculated in solve_vel should be based on the delta_v direction not the collision normal?

[Jondolf]

Thanks, I believe you're right. I probably just used the normal when I implemented restitution and forgot to change it to the delta_v direction when I added dynamic friction.

I'll fix this soon, probably as a part of fixing #2 if it ends up being related in any way.

[Jondolf]

You were right, it's supposed to use the normalized delta_v. Just using normal causes explosive friction in many cases, as can be seen in #47. Thanks for spotting the issue and sorry for fixing it so late 😅