I fully understand that there may be a commercial reticence to open up the world of off-the-shelf I2C products by giving us either an explicit I2C Bricklet, or an API on the IO-4/16. But frankly, I don't think it would undermine your sales of existing bricklets.. the beauty of the Tinkerforge product range is its extreme simplicity for non-electronics geeks. Most of these guys don't even know what I2C is. So IMO, providing I2C support is not going to cause people to suddenly not buy the "ready-made" bricklets because they can see ways of achieving the same functionality using I2C components sourced elsewhere.
Personally, I would think a new I2C API on IO-4/16 would be super-cool, and would show that Tinkerforge's philosophy is to allow its clients to truly tinker with whatever is out there, without locking us in to your product catalog. I also think that for Industrial customers, real I2C support would be extremely attractive.
Finally, I don't think it would hurt your marketing efforts if you could showcase a TF-based project that exploits the Pixy (and other I2C components).