If the image is accurate for color there can be a few things going on here. None easy to fix. As I had mentioned in an earlier answer about LED display technology, this is more than likely a metameric match produced my your measurement device from the narrow band LED backlight.

The problem with LED backlights (or any backlight for that matter) is that the measurement device must be able to decern the peaks and valleys of the spectral bands in the monitors backlight then integrate them to a standard value. Usually D50 or D65. But in reality the measurement devices see only pieces of the spectral bands, and in the case of the X-Rite CMUNDIS ColorMunki Display Calibrator it see's only what it's Custom designed RGB filter set provides. So if the RGB LED's don't fall perfectly into what the ColorMunki was designed for, it will miss a good portion of the colored light. That is more than likely what is happening here. TFT backlights are far better for color accuracy, but even with a full on spectro like the i1Pro2 getting a numerical white point match between two different display types rarely produces the exact same visual result.

It is not completely clear if the above is the issue, but If the calibration process completes and can be checked as passing, chances are that is the problem.

The other issue could be polarization. Both the measurement device and the monitor are polarized, so try turning the measurement device 45 degrees and re-calibrating the panel. If that doesn't produce vastly different results, choose another measurement device like an i1Pro2 to see if that is your problem. If not it is more than likely the technology used in the LED backlight. Tough stuff to calibrate and fraught with all kinds of metameric and CRI issues. Form my money it's TFT until that technical hurdle is leaped.

Adjusting my answer based on the updated information in the question: Since the camera has also recorded another more extreme cast, and the real case is quite subtle, then the issue could be simply that the closest XYZ value match to the aim white point on the Retinal display is not the same as the closest XYZ values to the aim white point on the LED display. To over come this you can use the set of targets I produced and posted here: Using this tool will allow those with patience, good hardware and software to adjust their displays removing the metamerisem caused by backlight differences. Have at it and have fun with it. :)