Another strong argument in favor of keeping Revenge is the lower rage cost + the (situational) cleave effect from talents.
Even if rage might not be an issue in most fights, you can still get those unlucky streaks of avoidance and/or simply have certain parts of a fight where you generate 0 rage - positioning when pulling, kiting, boss doing player based spells and not hitting you etc.
I have never really bothered trying to figure out how big a part the low cost Revenge plays in our rage management, but i am quite sure it's big. I mean, just compare the rage costs of Revenge vs. Devastate without factoring in SnB procs and the difference becomes clear.
A minimal dps and tps gain, but much lower tps/rage ratio or minimal dps and tps loss, but much higher tps/rage ratio?
"but threat is not a problem!" Sure, threat is not a problem in most fights. But rage does not = threat solely. Rage is also our fuel for various tools that we might need at critical points in a specific encounter.
200k self heal over the course of one fight assuming 6 healers is nothing. That's about one big healing cast each. That's one big heal you didn't stop in time. No big deal.
...but for 10 man raiding with 2-3 healers this might be a different story.
IV is situational. I will pur it into my second prot spec and try to exploit the living daylights out of it. If sucha thing is possible.
10 man is a totally different set of calculations.
Although the healing is worth more, your DPS is worth more as well. So while the healing might be worth more healing time, your DPS is worth more fight time as well. In fact, because DPS classes scale worse in 10-man compared to tanks, your DPS is probably worth a bit more in 10-mans than in 25-man.
So, the weighting is its own set of calculations really.