If you prefer, you can split MIDI and send both PROFILERs the same comands from Ableton via two seperate cables. This way both PROFILER s will perform the same transactions and be completely independent in case one "disappears". Then you don't need "UI to MIDI" at all. Mixing both approaches is not meaningful.
He needs both because in a performance situation he will be sending MIDI streams/instructions from Ableton *AND* "the player" will likely be manipulating the Kemper Master.
UI to MIDI would be taking care of "the player" changes, a MIDI merge or a dedicated splitter can take care of the Ableton instructions.
Ideally "the player" and the engineer can agree beforehand as to who would be manipulating what in order to help control data flow "conflicts", which in theory should be simply serialized by the MIDI system but in practice since the early 80's the implementation loves crashing your shyt face first.