just to let you know that I finally found the time to run a test with your fresh new boot loader + ubuntu image on kernel 4.4. It's muuuuuch faster than the initial tests. It's only 2% slower than what I got when I used to force the DDR to 792 MHz, so this is very good! It's really great to see this very good performance out of the box now! This board really is awesome for a server.
By the way, the method to enter maskrom by shorting the resistor didn't work at all for me. It used to work once by shorting the test pads however and I couldn't reproduce it. In the end I had to erase the beginning of the eMMC as suggested (don't forget to sync afterwards).
Also I think you should have mentionned that one of the biggest improvement with the new boot loader is the non-null wait time allowing to interrupt the boot and get a prompt. It will significantly ease testing of new kernels and images.