Validate that coincident HalfEdges are always congruent

[hannobraun]

HalfEdges that are coincident, i.e. located in the same space, must always be congruent. This means they must coincide exactly. The overlap must be complete. None of the coincident HalfEdges must overlap with just a section of another.

I discovered that the approximation code makes this assumption while working on #1605. It would be possible to fix this, but this assumption also enables some simplifications of the object graph. Which is why I've made the decision to confirm and document this limitation as a requirement, instead of lifting it. Given this decision, we should have validation code that upholds the requirement.

This might not be the right decision long-term, but it will work for now. And the simplifications it enables will certainly help us with any upcoming work. If it turns out that this requirement is too limiting, we can decide to lift it and add the required complexity to the kernel. But for now, I think this is the right call.

[hannobraun]

The issue description in missing some information on how to implement the validation. See this comment: #1697 (comment)

[hannobraun]

Closing in favor of #1937.