Validate that coordinate systems of curve definitions match

[hannobraun]

The curve that a HalfEdge is on is defined in local surface coordinates, through its path field. There are usually multiple such local definitions that refer to the same curve. HalfEdge refers to a Curve object (and as of this writing, still to a redundant GlobalEdge object, but that's on its way out; see #1937), to keep track which edges lie on the same curve.

There are validation checks to make sure that this all works properly, i.e. that coincident HalfEdges refer to the same Curve, and that HalfEdges that refer to the same curve are coincident. However, there is currently no validation check to make sure that the local path coordinate systems match. The path coordinate 0 of one edge might refer to a different point of the curve than the coordinate 0 on another edge.

This is turning out to be problematic (see #1968, for example; but this also affects my work on #1937). I think it's reasonable to expect all of those local coordinate systems to match each other, so each curve has a defined coordinate system. I'm not sure how difficult it would be to make this change, but a reasonable first step might be to enforce this through a validation check and see how much breaks.

Since this directly affects my work on #1937, I'm looking into this now.

[hannobraun]

I have a validation test for this in my local branch. I can confirm that pretty much all shells in existence are broken 😂

No idea yet how hard this is to fix, and I've run out of time for today.