I’m not a real embedded engineer.
I guess, MCU would boot from zero address and expect the bootloader here (like katapult), and this bootloader can boot things from 8kb offset.
If you flash Klipper without a bootloader, it should be flashed from the start (and compiled this way).
At least this is what I did with exactly the same chip: Cartographe V3: Reverse engineering
Hope that helps.