Add JAX 3D X-ray CT projector

[Michael-T-McCann]

See example results at https://scico--529.org.readthedocs.build/en/529/examples/ct_projector_comparison_3d.html Still not clear why the back projections don't look like they match. It seems like calling .transpose(2, 1, 0) on the ASTRA back projection gets them closer. My best guess is that this example generates vectors that cause ASTRA back projection to get flipped. I do think SCICO's back projection is correct given that it comes from autograd.

For consideration:

• Is the XRayTransform(Parallel3dProjector(in_shape, P, out_shape)) syntax too complex? I like that it separates the projector logic from SCICO (makes it easy to cut and paste into other projects), but it's awkward.
  • Removed XRayTransform

[codecov[bot]]

Codecov Report

Attention: Patch coverage is 27.02703% with 54 lines in your changes missing coverage. Please review.

Project coverage is 93.51%. Comparing base (d9026d3) to head (d182530). Report is 18 commits behind head on main.

:exclamation: Current head d182530 differs from pull request most recent head 308b602

Please upload reports for the commit 308b602 to get more accurate results.

Files	Patch %	Lines
scico/linop/xray/_xray.py	30.23%	30 Missing :warning:
scico/linop/xray/astra.py	16.67%	20 Missing :warning:
scico/examples.py	20.00%	4 Missing :warning:

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #529 +/- ## ========================================== - Coverage 94.67% 93.51% -1.16% ========================================== Files 91 90 -1 Lines 5707 5780 +73 ========================================== + Hits 5403 5405 +2 - Misses 304 375 +71 ``` | [Flag](https://app.codecov.io/gh/lanl/scico/pull/529/flags?src=pr&el=flags&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=lanl) | Coverage Δ | | |---|---|---| | [unittests](https://app.codecov.io/gh/lanl/scico/pull/529/flags?src=pr&el=flag&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=lanl) | `93.51% <27.03%> (-1.16%)` | :arrow_down: | Flags with carried forward coverage won't be shown. [Click here](https://docs.codecov.io/docs/carryforward-flags?utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=lanl#carryforward-flags-in-the-pull-request-comment) to find out more.

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[bwohlberg]

• Is the XRayTransform(Parallel3dProjector(in_shape, P, out_shape)) syntax too complex? I like that it separates the projector logic from SCICO (makes it easy to cut and paste into other projects), but it's awkward.

This is worth discussing. It may be worth hiding the Parallel3dProjector construction within the XRayTransform initializer. Maintaining the separation of the projector logic does seem worth maintaining, at least for now.

• Currently we have functions that covert from the SCICO projection specification to ASTRA one. Should we add functions that convert from a SCICO projection object to an ASTRA one? Where should these live?

Also worth discussing. The natural home for this would seem to be scico.linop.xray.astra.

• I find the relative positions of scico.linop.xray.astra.XRayTransform3D and scico.linop.XRayTransform in the module hierarchy a little odd. Time for a reorg (maybe after merging this)?

Is the concern that astra and svmbir are submodules of scico.linop.xray.astra? If so, I think this is justified by the view that the ones in xray are "standard", and the ones in xray.astra and xray.svmbir are additional options provided by other packages. But we can discuss this too. But I agree that we should revisit whether the "standard" transform should be imported from scico.linop.xray rather than scico.linop as is currently the case - it does seem a bit strange for the "standard" transform to have a home two levels up the module tree from the imported ones.