Why do self.set_camera_orientation() and self.move_camera() produce different results?

I'm trying to understand why `self.set_camera_orientation()` and `self.move_camera()` produce different final camera views, even when I pass exactly the same camera parameters (phi, theta, frame_center, zoom, etc.) in a ThreeDScene.

I expected both methods to end up with the same final camera position and orientation, with the only difference being that move_camera() animates the transition. However, the final views are actually different.

```
from manim import *

class Test(ThreeDScene):
def construct(self):
sp = Sphere()

# Uncomment this and comment out move_camera()
# self.set_camera_orientation(
# phi=80 * DEGREES,
# theta=-50 * DEGREES,
# frame_center=[0, 0, 3] - 3 * UR,
# zoom=0.9,
# )

self.add(sp)

self.move_camera(
phi=80 * DEGREES,
theta=-50 * DEGREES,
frame_center=[0, 0, 3] - 3 * UR,
zoom=0.9,
)

self.wait()
```
If I replace move_camera() with set_camera_orientation() using exactly the same parameters, I get a different final view of the scene. There's another confusing issue: when I use `self.move_camera()` and render with the -s flag, the saved final frame doesn't match the final frame I get when rendering without -s. I would have expected them to be identical

Could someone explain why this happens? Is this expected behavior, or is it a bug?

https://github.com/user-attachments/assets/51ce431d-a8ca-4b6b-abf3-5f31e9dc9762

Start by running the supplied ThreeDScene example with both set_camera_orientation() and move_camera(), including renders with and without the -s flag. Compare the implementations and final camera state after each path; done means the differing views are explained and the behavior is either made consistent or clearly documented as expected.

Written by the indexing model from the issue text.