Ok, so telling me that the intent was a 50/50 chance of one or the other result made the Tuaha conversationbundle a WHOLE lot more comprehensible. And here's the lowdown on how the inconsistency with Atsura happens.
The endslide considers the assassination successful if n_Quest_Tuaha_Success = 1.
Atsura's conversation says that Tuaha completed her mission if n_Quest_Tuaha_State = 3 ("Tuaha reports success and sends her thanks for your swift intervention.")., and fled if n_Quest_Tuaha_State = 4 (""Tuaha is safe now, but unfavorable circumstances forced her to depart sooner than planned.")
The 50/50 chance that you're talking about sets those two variables up in this way:
EITHER n_Quest_Tuaha_Success = 1 and n_Quest_Tuaha_State = 4, OR n_Quest_Tuaha_Success = 2 and n_Quest_Tuaha_State = 4.
If the Success is 1, endslide says the assassination happened, and if 2, it doesn't, BUT in both cases, because n_Quest_Tuaha_State = 4 regardless of the random factor, Atsura is always going to say "Tuaha is safe now, but unfavorable circumstances forced her to depart sooner than planned."
So for consistency with that dialogue, the 50/50 chance in Tuaha's dialogue should have set:
EITHER n_Quest_Tuaha_Success = 1 and n_Quest_Tuaha_State = 3, OR n_Quest_Tuaha_Success = 2 and n_Quest_Tuaha_State = 4.
Now what my mod does (I hope you can see why it looked like a bug and the Tuaha_State always being 4 meant the assassination should always fail if you didn't stop the stalker) is that it bases the endslide on the value of n_Quest_Tuaha_State, ignoring the value of n_Quest_Tuaha_Success.
I'll probably make an alternate version of my mod that does the fix I proposed above in Tuaha's dialogue, though that won't fix any save games where Tuaha's already been talked to. Would you agree that version would qualify as a bugfix? And I'll clearly label my original version a tweak, since it effectively removes the 50/50 chance.