Is natural light ever circularly or linearly polarized? I understand that all light is elliptically polarized, with circular polarity being a special case in which the quadrature phases are offset exactly 45 degrees, and linear polarity being an even more special case in which the propagation is limited to a plane.
Is black body radiation polarity constrained beyond an ellipse?
Or, equivalently: Is the light we observe in nature – emitted by the sun or incandescence and transmitted through our atmosphere and reflected by natural opaque objects – ever strictly circularly polarized?
 A: I think you may have taken something you read a bit too literally.
Natural light, like direct sunlight, is generally randomly polarized. If you could measure the polarization at an instant, in the next instant you would find that its polarization is randomly different.
Your question seems aimed at trying to understand polarization that occurs in nature, from a photographer's point of view.  The important thing there is that even randomly polarized light can be separated into orthogonal components (up/down vs right/left).  When light strikes a glossy surface at a fairly shallow angle (~57 degrees from perpendicular), practically all of the light components whose polarization is mostly perpendicular to the surface are transmitted through the surface and scattered or absorbed.  The light components whose polarization is parallel to the surface are nearly 100% reflected.  So light reflecting at an angle off of water is linearly polarized in a horizontal direction.  If you were under the water surface looking up at the right angle, the light you see would be mostly linearly polarized in the vertical angle (that is, in a plane that is perpendicular to the water surface).  A linear filter serves to remove reflections in windows for precisely that reason; and it changes the color and contrast in photos of lakes and rivers for that same reason.  Fly fishermen often wear linearly polarizing filters to block horizontally polarized light so that they can see down into the water better.  That effect is due to Brewster's Angle.
Circularly polarized light occurs in some fun situations due to natural light.  No doubt you've seen colored reflections off a tabletop when there was a CD case or other plastic object on the tabletop.  Those are due to a combination of linear polarization due to the first surface the light enters, then the plastic converts the linear polarization to elliptical polarization depending on the thickness of the plastic and its amount and orientation of birefringence (usually due to stretching or flow during manufacture), and then another round of linear polarization filtering by the tabletop.  Some plants and insects have surfaces that act as circular polarization filters for reasons I'm not equipped to explain.
A circular polarizing filter is usually a birefringent layer that converts circular polarized light of one chirality into linearly polarized light, then a linear polarizer to select out just the linear portion of that light in one orientation (vertical or horizontal relative to the birefringent axes of the first layer).
The distinctions between linearly polarized light, elliptically polarized light, and circularly polarized light can be confusing.  I'll try to make them clear.
Let's start with linearly polarized light, with the polarization axis vertical.  Now tilt that axis 45 degrees so that it reaches diagonally across a square whose sides are horizontal and vertical.  If you place a polarizing filter in the beam, with the "pass" axis vertical, you will get light through corresponding to the vertical height of the square.  Rotate the filter 90 degrees and you will get light through corresponding to the horizontal width of the square.  You've just decomposed the diagonally linear polarized light into vertical and horizontal linearly polarized components.
You can't see it directly of course, but the vertical and horizontal components are precisely in phase with each other at this point in the experiment.
Now take out the polarizing filter and put a birefringent filter in the beam, with its axis vertical (the beam still has its 45 degree polarization tilt).  Light going through the filter with vertical polarization will be delayed a bit relative to light going through with horizontal polarization.  If the delay is precisely enough to retard the phase of the vertically polarized beam by 1/4 of a cycle, the beam will emerge with right-circular polarization.  If it retards the phase 3/4 of a cycle, the beam will emerge with left-circular polarization.  If the amount of retardation is 0 or 1 cycle, the linear polarization is unchanged.  If the amount of retardation is 1/2 cycle, the polarization remains linear but gets flipped 90 degrees (the opposite diagonal of the square).  BUT if the retardation is anything between those values, the beam will emerge elliptically polarized.
