[WIP] Naive off-axis slice attempt for spherical-ish geometry

[matthewturk]

This is an attempt at addressing #4750. It works alright for offsets in theta, which seem to just rotate, but I'm confusing myself when attempting to test it for having a new normal vector that is set at some declination.

I think I've also confused myself yet again about how the r, theta, phi coordinates work, as my test script is not producing the types of plots I'd expect. When slicing along the dimension that runs 0 .. pi, I'd expect to get a whole circle, and when slicing along 0 .. 2pi I'd expect half circle. But that's not what's happening.

Here's an example script:

import numpy as np
import yt, yt.testing

ds = yt.testing.fake_amr_ds(geometry = "spherical")

for i, phi_offset in enumerate(np.mgrid[ds.domain_left_edge[2]:ds.domain_right_edge[2]:100j]):
    s = ds.slice("theta", ds.domain_center[1], offset = [0.0, 0.0, phi_offset])
    s.to_pw("Density").save(f"frame_{i:04d}")
    print(i, phi_offset)

[matthewturk]

There are lots of problems with this! I am rethinking it but it was absolutely helpful to sketch it out.

For starters, it's incorrect - an offset isn't that same as a new north vector. But I do see a path forward here.

[matthewturk]

Moving to draft, and I think I've found the right math now and will revisit.

[matthewturk]

This is going to stay WIP, but I've been poking at it a bit and will try to come back to it very soon.