Add point-circle centrality constraint

[tomcur]

On top of #68.

This adds a point-circle centrality constraint, reusing the expressions introduced for point-point coincidence.

[tomcur]

Note that this does not introduce any new expressions, it's conceptually very similar to PointPointCoincidence.

[tomcur]

(Also having valency 2.)

[waywardmonkeys]

If we keep referencing, then this is the same as point-point-coincident constraint! (And I think we should keep referencing as we've been discussing on a call.)

[tomcur]

If we keep referencing, then this is the same as point-point-coincident constraint! (And I think we should keep referencing as we've been discussing on a call.)

Indeed, I'll convert this to draft for now; and if we keep referencing, we should probably close this as to not have too many ways of doing the same thing.

For posterity, I'll paste part of our thinking on referencing (and related topics) posted elsewhere here:

We're thinking more about constraint graph analysis, and how that'd work with referencing. It's especially the degree-of-freedom counting that is complicated, as, say, a boundless line in 2D has two degrees of freedom, even if it is defined by referencing two points that have four degrees of freedom in total. Constraints can be placed on those points and the line separately, which needs to be held into account when analyzing the system. Then, separately, you may have constraints that take away specific types of degrees of freedom (like a constraint on a circle's circumference taking away the degree of freedom of the radius, and not the location, which ideally you detect).