Dev/coordinate transforms

[austinschneider]

In this PR we are trying to more strictly enforce which coordinate system is used in which part of the code. The main upshot is that coordinate transformation bugs in the distributions project will be much much much harder to make.

We are going about this by having the "Earth coordinates" only be used internally by the DetectorModel class.

This is implemented in two ways:

1. The public interface of the DetectorModel class only uses the "Detector coordinates", with few exceptions.
2. All methods of the DetectorModel class that accept coordinates of some form use named "strong types" (implemented with fluent::NamedType<T>) for dispatch, so that the particular method being called is clear at the invocation.

There are three exceptions to this division of the coordinate systems:

1. The IntersectionList still contains "Earth coordinates"
2. The Path class internally uses "Earth coordinates"
3. The DetectorModel has methods for setting and getting the detector origin, which by construction must be in "Earth coordinates"

The first two exceptions exist so that we can avoid copious coordinate transformations in regular usage of the Path and DetectorModel classes, and are fine since we do not expect users to ever interact with the IntersectionsList internals. A consequence of this is that the SamplePairProduction method of the Injector class had to be removed (which was going to happen for other reasons in any case).

Ideally we do not want users on the python side to have to think about two coordinate systems, so the pybindings will be changed to only use the parts of the interface that accept DetectorPositions and DetectorDirections (with the exception of getting or setting the detector origin), and to perform the conversion from Vector3D to these tagged types automatically.

This PR may also solve the issue presented in #44
I have already come across several instances where the coordinate transformations were not correctly applied within the distributions project.

[austinschneider]

The latest commits fix an issue in the DIS cross section that was causing generated events to look like they fell below the Q^2 threshold.

[austinschneider]

This should now sort out all the issues I was having with Alex's ATLAS example script.

[nickkamp1]

The last commits get the DarkNews examples working again (as in, they generate events and there don't seem to be any coordinate system mismatches). Borrows some commits from the feature/examples branch. I think this branch is now ready to merge into main