I would agree with you that we're looking at a shiny (bright, high reflectivity) surface, but only in certain wavelengths, at least when compared to the sky. You referenced the navcam image which shows a darkened sky, and there are many pancams that show the same effect, but it is only those taken through certain wavelengths that show the effect.

The navcam is infact filtered, by three filters in fact, to obtain a red-band pass filter centered at ~650 nm. The exact spectrum is available at , page 6 :
http://robotics.jpl.nasa.gov/people/rwillson/papers/2003JE002077.pdf

When looking at the spectrum, navcam is apparently equivalent to combining pancam filters L4, L3 L2 at a ratio of about 5:3:1

There are a great many pancam images that show the same effect for Spirit, but more so now than when the mission began. I'm speculating that this might be an indication of the dust composition having lowered since the mission began as mentioned in the msss weather reports for the landing sites ( http://www.msss.com/mars_images/moc/mer_weather/weather/15.21.May.04.MER.html ), creating 2 effects: First, there are less dust sized particles capable of scattering red light and second, there is less overall brightness to the opposition sky.

This effect is showing up most prominantly in afternoon sessions pointing towards the Columbia hills, on Spirit. The Columbia hills lie East of the rover, and as you described, the ground might be showing the heiligenschein effect, which was noticed and described in the dry environment of the Moon (its prominance in that environment owes more to the lack of sky-brightness / scattered light than to light being focused by water droplets as was mentioned in the article you linked.)

This "Opposition Effect" in lunar photography is described at http://www.sundog.clara.co.uk/atoptics/oppos1.htm

So what does this tell us about the sky color? Its almost impossible to tell a sky color from the images as have yet been released, as there is no way to know how the brightness of an L3 (red) image relates to an L6 (blue) image, without more knowledge about the exposures of the images themselves. What we can say is that the brightness surface features of red images relative to the sky is high, and that blue images still show the opposite (sky brighter than ground) effect.

The one remaining mystery to me is the direction of the brightness gradient in the 'dark sky' images. It is more noticeable in the pancam images, but the gradient of brightness in red filtered images goes up from the horizon to the higher altitudes.

This series of spirit Sol 149 show the (highly exagerated, by long exposures times on the green and blue images) effect:
(i apologize for raw links, i haven't mastered imbedding images or links here yet)
L7 (blue, lightsky)
http://www.lyle.org/mars/imagery/2P139592958EFF6104P2272L7M1.JPG.html
L2 (near infrared, darksky)
http://www.lyle.org/mars/imagery/2P139593029EFF6104P2272L2M1.JPG.html
L3 (red, darksky)
http://www.lyle.org/mars/imagery/2P139593307EFF6104P2272L3M1.JPG.html
L4 (orange, darksky)
http://www.lyle.org/mars/imagery/2P139593389EFF6104P2272L4M1.JPG.html
L5 (yellow/green, lightsky)
http://www.lyle.org/mars/imagery/2P139593488EFF6104P2272L5M1.JPG.html
L6 (green/blue, lightsky)
http://www.lyle.org/mars/imagery/2P139593708EFF6104P2272L6M1.JPG.html

A color composite of these images, unadjusted, led to one of the brightest blue skies (the sky brightness and color is most likely due more to elongated exposures of the yellow/green/blue images, as witnessed by the bright blue features in the rocks), but the effect is very striking representation of the variation of color between different parts of the sky: