I understand they're actually multiple greyscale images, that's just how you get CP to handle multichannel TIFs correctly.
Your TIF didn't work for me at first but when I opened it in FIJI (a type of ImageJ) I saw that the .tif you sent had the 4 images marked as 4 z planes, not 4 channels- I ran Hyperstack->StackToHyperstack, changed channels to 4 and z planes to 1, saved the new image (attached here). Then it worked perfectly.
Use the image I've attached to make sure you've got your pipeline configured correctly, then try your .flex image in that same pipeline and see if it passes or fails. If it turns out your images won't work as configured, I can send you the code for an ImageJ/FIJI macro that will just go through and split all your images into their 4 channels.
flexsplit_channels.zip (4.7 MB)