You're right! I didn't know this excited but it turns out that RealD glasses use circularly polarised light. This observation would have been impossible with linearly polerised light as symmetry prohibits 90° rotations (+90 and -90 cancel). If we assume that the mirror indeed turns right into...
Over each eye is a linear polariser and they are orthogonal to each other. So I conclude from the experiment that the polarisation must have been rotated by 90° by the mirror.
That reminded my of phase plates but they only work because their refractive indix depends on the angle. Along two...