Ok you are right Donald. An interrupt could occur at that moment between fetching/executing the instruction that attempts to place the processor into the idle state. If there is an interrupt that comes at this point it will happen irregardless as near as I know. If it comes just before the idle mode is "official" the processor will go to the interrupt right away and when complete after its RETI instruction will come back to the JMP to TOP_MAINLOOP instruction which follows. If the IDLE has become "official" and then the interrupt comes it will immediately quit the idle mode and once again respond to the interrupt and do the RETI and come back to the JMP TOP_MAINLOOP.

I do recall a long long time ago (10 years or more) reading something where it was recommended to put one or two NOP instructions between the IDLE setting instruction and the JMP TOP_MAINLOOP instruction. Why this was even ever necessary is still unknown to me.

Michael Karas