To explain what you're looking at:  Those ellipses are actually outlines of spheres that have been scaled down on one or more axes so that they are elliptical spheroids.  If you want to, you can select each ellipse (red and green) and set the Element Display Type to wireframe, and then you can see them in 3d, but it wasn't important for this test and I didn't want to clutter the procedure with non-essential steps.

As to what those spheroids mean, they are envelopes that describe the falloff of a given joint's influence of bend; within the green falloff zone, "full bend" will be applied; outside of the red falloff zone, "no bend" (or "no twist" or "no side-side" depending on the axis) will be applied; within the red zone, but outside of the green zone, "mixed bend" will be applied.  If these falloff zones are not symmetrical, then the deformation applied by bending will not be symmetrical.  This is bad (hopefully for obvious reasons).

There are other aspects of symmetry that are borked, but this is the reproducible flavor; it may be that some or all of the other problems with Poser's rigging (asymmetry at zero pose, or the tumbling falloff zone bug) are all aspects of this same bug.  I appreciate the time you spent checking this, thank you.

Oops, one thing more: no matter what any joint's parameters might be set to, if you do Figure -> Symmetry as described earlier, this should force them to be symmetrical.  It doesn't, and it may be that the Symmetry tool has a problem, but (maddeningly) even if you hand-edit the CR2 file and supply symmetrical values, Poser still reads the file wrong (apparently) and leaves the rig in an asymmetrical state.  Very bad bug.

My Freebies