First pass at winding in the worldine formulation

[evanberkowitz]

This is a PR that should ultimately implement the Winding_Winding and WindingSquared observables.

HOWEVER, it should not yet be merged. There is a big issue though:

If δ/δJ (dδJ) = 0 then the worldline WindingSquared is negative semidefinite and the Villain is positive semidefinite. If there is no mistake this would prove that WindingSquared = 0, which

• is false from running Villain samples.
• would imply that very plaquette's winding = 0, also false.

[evanberkowitz]

With these edits the WindingSquared observable now matches between actions

and so does the Winding_Winding observable

(blue = Villain, orange = Worldline, x-axis: √(∆t^2 + ∆x^2), y-axis Winding_Winding)

The matching is sufficient to believe the observable is error-free.

[evanberkowitz]

Closed #29.