Resolution is the key
You missed the real problem with colour eInk - resolution.
A colour eInk display requires 4 separate cells (RGBW*) for each pixel, and each cell needs to have at least 4bits of scaling to produce marginally acceptable colour. Now, current eInk displays on the market have 4bit displays at a res of 166dpi for a 6" screen (200dpi for 5", 150dpi or less for larger sizes). Divide that by half for the 4-cell pixels and you get 83ppi, about 30% less resolution than current LCDs and far too little for decent text at 12pt. The 166dpi of current eInk readers is 'good enough' to display small text when combined with greyscale aliasing, but going down to half the res would produce a pixellated mess that isn't going to sell.
So, the big challenge is to increase the resolution of the screens, something which requires a major upgrade to manufacturing and quality control techniques that no-one's cracked yet. I'm sure we'll see low-res colour epaper screens used for point-of-sale and other sorts of displays, but it's a while still before someone will come out with a colour device that will be comfortable to read for long periods while a foot away from your nose.
*As for the filters, they're RGB because the actual reflective surfaces are either white or black and the filters lie over them, so you need to use transmissive filters which shine red light on the white balls for a red dot, etc. Ink on paper works differently since the ink only reflects part of the spectrum, whereas eInk balls either reflect all of it or none. If they were able to create cells with different pigments then they'd be using CMYK, but that would need some very costly manufacturing. The W is a white, or clear cell that's needed to increase the contrast.