Update getBodiesUnsafe() to return a slice of size getMaxBodies()

[r0ckHopper]

Currently getBodiesUnsafe() returns a slice of size getNumBodies() but getBodiesUnsafe() actually returns a pointer to slice of size max_bodies.

The issue comes when you delete bodies from the middle and try to access bodies of higher index

getBodiesUnsafe does not reorganize the elements before giving a slice so you can get an out of bounds error

example of the issue is:

let bodies be
[b0, b1, b2]

if you delete b1 jolt physics updates bodies to
[b0, invalid, b2]

but if you call getBodiesUnsafe you only get
[b0, invalid]

so if you do tryGetBody( all_bodies, b2) then it will look for b2 at index 2 which is out of bounds

[Copilot]

Pull request overview

This PR fixes an API mismatch where PhysicsSystem.getBodiesUnsafe() / getBodiesMutUnsafe() returned a slice sized to getNumBodies() even though the underlying Jolt buffer is sized to max_bodies, which could cause out-of-bounds indexing when accessing bodies by BodyId after deletions create holes.

Changes:

• Return a getMaxBodies()-sized slice from getBodiesUnsafe() and getBodiesMutUnsafe().
• Add a regression test covering the “destroy middle body then index by BodyId” scenario.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[r0ckHopper]

@hazeycode I think my response to the comment will need human input.
Zig doesn't have a clean try catch for panic and copilot's suggestion would not be testing the issue. Current options are to
1.Keep test as it is and test only happy path
2.Create a subprocess and check exit code (this might be a bit overkill for this project's test code)
3.Delete the test

[hazeycode]

Yeah it's fine as it is for now.

[r0ckHopper]

Updates the the doc comments as well. Should be good to merge now unless you wanna tweak the wording.