Ed
I just read your document. I am surprised that the TFTP loader is so large - the uTasker project has ARP/ICMP/IP/TCP/HTTP web server interface loader which is 16k in size on the Kinetis devices. I would estimate TFTP to be around 10k (although not knowing what else may be packed in there).
The uTasker serial loader has SREC, USB-MSD, USB-HID (KBOOT compatible), (encrypted) SD-card and Ethernet web server loaders and works on all KL and K devices (when the required peripheral interface is available in the chip) and can be built with KDS, CW, Rowley, GCC, IAR, Keil, Atollic, CooCox or VisualStudio (for real-time simulation).
Have you considered dropping your code in to a task in there and making use of the integrated TCP/IP resouces, file systems, etc. as well as real-time simulation/development/debug capabilily since it would then add a uMon-M mode to it that will then be compatible with all the devices and IDEs?
Regards
Mark
http://www.utasker.com/kinetis.html