Well, if we're gonna go with dominant > OU > Banned > BL > Non-existent > UU, it's trivial to make at least a preliminary ranking. I drafted this up based on the Smogon tiers (and I know that especially for BL they don't exactly fit what we're talking about, but I'm sure we can hash that out later) and on Jorgen's list above of the top Pokémon in each gen. I counted Jolteon as BL in gen 1 and Raikou as dominant in gen 2.
If being dominant is worth three times as much as just being OU, and all other states being worth close to zero, we get:
10: Tyranitar (3 dominant, 1 OU, 1 non-existent)
~8: Snorlax (1 super-dominant, 1 dominant, 2 OU, 1 unused)
7: Jirachi (2 dominant, 1 OU, 2 non-existent)
Gengar/Starmie (1 dominant, 4 OU)
6: Zapdos (1 dominant, 3 OU, 1 unused)
5: Celebi (1 dominant, 2 OU, 1 banned, 1 non-existent)
Suicune (1 dominant, 2 OU, 1 non-existent, 1 unused)
4: Heatran/Rotom-A (1 dominant, 1 OU, 3 non-existent)
Raikou (1 dominant, 1 OU, 1 BL, 1 unused, 1 non-existent)
Exeggutor (1 dominant, 1 OU, 1 BL, 2 unused)
Jolteon (4 OU, 1 BL)
Blissey/Forretress/Skarmory (4 OU, 1 non-existent)
Cloyster (4 OU, 1 unused)
3: Chansey (1 dominant, 3 BL, 1 unused)
Garchomp (1 dominant, 1 banned, 3 non-existent)
Keldeo (1 dominant, 4 non-existent)
Politoed (1 dominant, 3 unused, 1 non-existent)
Tauros (1 dominant, 1 BL, 3 unused)
Dragonite (3 OU, 2 BL)
Metagross (3 OU, 2 non-existent)
Gyarados/Vaporeon (3 OU, 2 unused)
This is pretty close to what was being suggested already.
I like the idea of further dividing OU into Pokémon that actually do something in the tier (even if they aren't dominant) versus those who were just there. I think part of what feels "dominant" about a Pokémon across generations is that it imposes itself. To that end, I feel that the difference between, say, gen 1 Starmie and gen 1 Persian is bigger than the difference between gen 1 Persian and gen 1 Gyarados, even though the first two are OU and Gyarados is UU.