Use lazy iterator to ensure all rings in create_[a/b]_list are closed

[asinghvi17]

Fix #191 by ensuring that an extra point is effectively "vcat"'ed onto the stack, in case the polygon ring is not closed.

At some point this will change to use whatever struct we implement for the lazy polygon winding order / lazy polygon closed ring stuff, but for now this seems fit for purpose.

[asinghvi17]

Ah, this breaks cut output because cut relies on the second ring not being closed. Maybe we have a pass through wrapper DontFix{T} for situations like this one?

I could also make the lazy closed ring wrapper and wrap everything in intersection, difference, and union in that. In retrospect this seems like a better plan.

[rafaqz]

We may need a dedicated wrapper type for this to be fast.

[asinghvi17]

Tests fail because this doesn't handle empty rings (I guess that it should be an empty ring in that case...)