I would think the first thing would be to set PhysicScale() to the right value, since everything is dependent on that. I have settled with setPhysicsScale(0.2) which to my understanding, should give me precision on the pixel level. (don't even remember how I got to that)
Still, I'm clueless about most of the values. I can't even figure out what the "right" mass would be. Could be 0.000001 or 1.0 or 100000. And since all those values are dependent on each other, I quickly end up jumping between them, hoping to find something that just works. One thing I figured out is that raising SetSpritePhysicsAngularDamping() can stop my sprite from being stuck in eternal spinning.

Are there any way to get those values to make sense, other than throwing random numbers at them and hoping that something sticks?

I can't even figure out what the "right" mass would be.

weight/gravity = mass

The thing i found out was you need to have a very small virtual resolution in order to get any decent physics speed.
All my tests seemed to have the velocity top out when the virtual resolution was closer to the window size.
You can get the mass with GetSpritePhysicsMass() (There are quite a few physics functions)

I wrote a fancy dev function enabling me to edit the values in real time. No dice. The sprite still went totally batshit.

But then I finally figured out that ... I made a mistake. I had forgotten to normalize the rotation to something within the 0-360 range. Once fixed, everything just works.

Thanks for the help guys!