From what's been hinted at there's already a lot of progress for the pi4, in terms of thagrols list of steps (taking it in a more generic form rather than eMMC specific) many of those steps have probably been ticked off already (potentially after several itterations). Each of those steps will have a price associated as well as time.
As W.H.Heydt said above, there has been progress with each itteration of the models and that is expected to continue with the pi4 and pi5 ranges (it's also been hinted that the Engineering team have already started on the pi5 - although that may only be some early designs/specifications).
Going back to the question of eMMC I've got some boards with both eMMC and raw NAND chips, whilst it's nice having storage integrated it does increase the complexity and also adds other issues. The boards with raw NAND are particularly problematic as they don't have another way to easily boot and they're much more reliant on the OS doing the right thing. For the most part I stick with my raspberries as I know if I plug in a good sd card they'll just work and keep going with regular updates, for the others it takes more effort and skill from myself to keep things going (last time I tried one of the eMMC boards it needed fixes in the kernel to enable eMMC)