One thing I noticed about the legends in the live-zombie-leader save I have is that the tower appeared to be doing a lot of attacks before they got to the player, and the leader was generated on one of those attacks (against a forest retreat). So it might be tricky capturing the live leader being created in the act.
So necromancers are dying in battle and being replaced by mortals, ala lieutenants and generals, but it's not being properly recorded? I'm confused.
It occurs to me that some of trouble here may be to due the unusual entity status of necro towers.They're site-level governments, but lack parent civs and are militarily active.
My understanding of what Toady said is that the siege code decides it needs a leader when a siege actually happens, and the necros aren't (always?) selected as leaders. In the save Toady got from me the necro left more or less immediately, leaving the leader/handler and the undead to (mis)manage themselves. My experience is that necros tend to leave rapidly, with the only exception being stealthed ones who actually try to sneak in (fleeing when detected, and caught in cages when not).
I think the save Toady needs to investigate that issue further is one where the necro tower(s) is isolated from everyone except the player fortress with a save just before they (frequently) show up, and when they do, they usually have a leader/handler with them. It should probably be before the first siege (from that tower), or at least previous ones should not have included any leader/handler. Not the easiest thing to produce, but might be attempted at the same time as hello/goodbye sieges are attempted to be reproduced.