Fix world interpolation

[Markonis]

This PR simplifies the stepping logic, and fixes a bug with jerky interpolation.

[DakotaLarson]

Hey @Markonis, thanks for contributing! I am looking to probably add this logic to my own fork.

Can you provide additional context around the scenarios the initial commit is addressing? I have had difficulty with "missed collisions" in the past that I addressed by reducing dt to 1 / 120 when calling World.fixedStep and was wondering if this makes improvement to that process.

[Markonis]

Hey @DakotaLarson sorry for the delayed reply... The issue that is being addressed is the following.

Because the physics world has a fixed step, it can happen that the physics FPS is actually lower than the renderer FPS. In that case, for each renderer frame we need to interpolate between 2 frames of the physics step, to make the movement appear smooth. But here's the crux of the issue: for the interpolation to be correct, the current renderer frame needs to always be between a physics frame that was in the past and the one that's in the future.

Renderer frames: R---R---R---R---
Physics frames:  P------P------P---

So if you look at the last 2 R frames, they both fall between the last 2 P frames. But if we get the next R frame, like so:

Renderer frames: R---R---R---R---R
Physics frames:  P------P------P----

We need to ensure that we compute the next P frame that's in the future beyond the last R. This wasn't happening previously...

[aziyatali]

We should enable branch protection