After some further thinking, I see the biggest issue being debugging a kernel without any form of display.
EDIT: They do have a serial console, along with JTAG headers as discovered
here.
This means that the first thing that will have to be written is some form of TTY console driver for the e-ink panel (which means we need to find the panel datasheet/spec somehow).
As for testing, there are tools to let us boot a kernel from inside an running linux system, which means that we wouldn't have to flash to the memory every time, but boot it from data stored on a sd-card.