The G load increase is comparatively slow, but the release is pretty sudden.

Lots of things "bounce" in funny ways when you put them under very large loads, then release that load suddenly.

OK, so my knowledge of physics is lacking, but wouldn't the decrease of non-centrifugal G-load also be a bit slow as it takes the atmosphere some time to slow down the speed of the projectile?

The radial load/unloading is (approximately) instantaneous at the moment of release and does not vary in proportion to wind resistance.

Edit: this is, btw- fixable. Like other commenters mentioned, a trebuchet/sling like release mechanism would slow that radial unloading and reduce stress on the spinning arm too. It’s a lot more complicated to achieve though (as if this prototype/trial weren’t complicated enough!)

During spin-up the lateral force on the projectile increases slowly until it reaches maximum. When it's released, this load disappears almost instantaneously and the object is effectively in free-fall ... effectively 0g.

Then it hits the atmosphere and is subjected to significant drag, but that onset is reasonably slow (by some metric) compared with the release mentioned above, but it's along a different axis. What's more, the projectile has spin along its long axis ... I don't know how that's imparted.

Yeah, there are some huge forces that change rapidly in lots of directions.