From plain looking at your code
(without really understanding UEFI stuff)
I see nothing what could cause a regression.

Therefore I simply accept it as a good starting point
towards USB UEFI boot support #831

In particular I like it because that all the actual work
for USB UEFI boot support happens in one single
separated script output/USB/Linux-i386/10_create_efiboot.sh

In that script regarding

# Should we crash if umount fails?
umount ${EFI_MPT}

I checked what the rear *.sh scripts do in general
when umounting via

$ find . -name '*.sh' | xargs grep -A3 'umount ' | less

and in almost all cases the exit code of
a plain "umount" command is ignored
except in lib/global-functions.sh

Interestingly the umount_mountpoint function
in lib/global-functions.sh is nowhere used ;-)

Thanks for the feedback, It was important for me to know that the starting point is right. Now I can continue to work on it.