I have no idea whether the cadence is an indication of anything suspicious but I'm surprised no one has bothered to check it since there's a lot of guessing going on.
I have a PAL video recording of the stage. PAL video is recorded at 25 frames per second, which means that there is 40 ms between each frame. I ran the video in VirtualDubMod and cheked the frames when Froome made his attack. I've grabbed a screenshot each time his left knee is at it's upmost position:
You can see the frame numbers and timings below the screenshots. He makes three full rotations within just 37 frames (1.48 seconds). Hence the cadence is 3/1.48*60 = 121.6
Regarding the accuracy: If we consider the pedal stroke as a 360 degrees rotation, then the feet (pedals) will have moved 30 degrees between each video frame at such a high cadence. This means that the pedals will not necessarily be at the exact same position whenever a rotation has occurred but somewhere within 30 degrees. Therefore analyzing a longer stretch is more reliable. But the error is +/- one frame (40 ms) for the entire sequence, which means that the cadence at this very moment is between 3/1.52*60 = 118.4 and 3/1.44*60 = 125.
Froome flies away from Contador and catches Quintana in just over 100 seconds, and goes right past Quintana - still with a cadence of 115!