Another change is in the firmware config.txt:
gpu_mem=32 # set ARM to 480Mb DRAM, VC to 32Mb DRAM
framebuffer_ignore_alpha=1 # Ignore the alpha channel for Windows.
framebuffer_swap=1 # Set the frame buffer to be Windows BGR compatible.
disable_overscan=1 # Disable overscan
init_uart_clock=16000000 # Set UART clock to 16Mhz
hdmi_group=2 # Use VESA Display Mode Timing over CEA
arm_freq=900
arm_freq_min=900
force_turbo=1
So they are setting a fixed clock now. I know that from another company who made a HAT running on Linux with many sensors, who found was necessary to force turbo to get rid of sync issues with some SPI and I2C chips. So Microsoft seem to be moving forwards with their testing/stability (finding and solving the same issues).
Great progress anyway and plenty of teasers for what is to come

I hope the deploy/debug issues with different versions can be solved quickly, because they must be stable first to increase the speed of both their remaining development and our "maker" development which justifies the whole project. Looking forward to next week...
p.s. it's probably a good idea to copy those last few lines into your RTM config.txt to minimize issues in the meantime. Personally I also set hdmi_group to 1 which solves initialization problems with some external monitors (I use a 7" FPV monitor).