Make code easier to read.

[bangerth]

The current code does

  Object x;
  sometype &y = x.internal_representation();
  y.a = ...;
  y.b = ...;
  return x;

This is hard to read, primarily because to the reader, the association between x and y is lost. But it's easy to avoid this.

Follow-up to #264. @alarshi FYI.

[github-actions]

Benchmark	Master	Feature	Difference (99.9% CI)
Slab interpolation simple none	1.216 ± 0.004 (s=374)	1.217 ± 0.004 (s=368)	+0.0% .. +0.2%
Slab interpolation curved simple none	1.290 ± 0.067 (s=355)	1.285 ± 0.066 (s=346)	-1.7% .. +0.9%
Spherical slab interpolation simple none	1.348 ± 0.052 (s=351)	1.339 ± 0.055 (s=321)	-1.7% .. +0.3%
Slab interpolation simple curved CMS	2.048 ± 0.019 (s=210)	2.042 ± 0.018 (s=233)	-0.5% .. +0.0%
Spherical slab interpolation simple CMS	2.241 ± 0.073 (s=201)	2.218 ± 0.081 (s=205)	-2.1% .. +0.1%
Spherical fault interpolation simple none	1.353 ± 0.116 (s=334)	1.344 ± 0.122 (s=336)	-3.0% .. +1.5%

[MFraters]

I like the simplification. Although it looks the correct to me (same code, just some reorganization), for some reason it fails the debug tests, complaining about the following for the spherical tests:
what(): Assert `boundary_points.first[i] <= boundary_points.second[i]` failed in /home/runner/work/WorldBuilder/WorldBuilder/include/world_builder/bounding_box.h at line 264: Bounding Box can't be created: the points' order should be bottom left, top right!
for two spherical tests: 24 - spherical_fault_interpolation_simple_none and 25 - spherical_slab_interpolation_simple_CMS.

This could either indicate that there is some small mistake I am missing or that there may be a bug in the original code. Could you have another quick look at it to see if you can find it?