aramis's math notwithstanding, I suspect that in all of the games you mention, the GM is responsible for NPCs attacking PCs. Ergo, each of your examples cease to be deadly when the GM decides that the NPCs are done fighting.
Which said math is only about the need for retainers; I should have been more clear.
I'll note, however, that your assertion is problematic, given that a not insignificant minority treat many RPGs as having fixed rules that prescribe a certain setting by mechanics. In the case of D&D (at least, BX, AD&D1E, BECMI, AD&D 2E, Cyclopedia, and optional in 5thZ), there's a rule for...
- Determining if an encounter happens (Random Encounters rule)
- Determining what is encountered (Random Encounters rule)
- Determining whether or not the encountered being is hostile (Reaction Rule)
- Determining when they should withdraw and/or surrender (morale rule)
For those transitioning from boardgames with fixed rules, that the GM is to moderate this is inobvious, even given the table building aspect of random encounters in AD&D 1E/2E. Every monster has a number encountered entry.
For the rules-stickler, on the fly rejection invalidates the point of rules; adapting the rules for setting usually doesn't. One of the reasons I quit being a player was that most other GMs ignored rules willy-nilly, often with no thought about knock-on consequences. For such GMs and/or players, the decisions about deadliness are made at the table creation stage, not the in-play running-the-encounter stage.
I happen to have tended towards that mindset most of my life... so the lethality is determined in large part by the rules.