• with algorithm changes in feat(developer): double markers once again 🙀 #10541 feat(core): ldml double marker, C++ side 🙀 #10563, we actually don't need the boundaryBefore code here. So feat(developer): ldml load ccc from icu 🙀  #10515 is obsolete.
• the NFD algorithm won't reorder text over a normalization-safe boundary. so the segments are still 'there' but they are ICU's problem.
• We may need the boundary detection in core for minimizing context changes, but it's a first class function in ICU so doesn't need anything in Keyman and definitely not in kmc.

#10317

@keymanapp-test-bot skip