There isn't a 25% chance of the Ferrothorn dying unless the Genesect user is actually a banana, and U-Turning out of it if you don't have something else to handle Ferrothorn (like Celebi) is a terrible idea. Going for the kill when the opponent expects a momentum grab is an incredibly viable strategy that has nabbed me an opening 6-5 in a great number of matches (and left me on the bad end in a few...). just because prediction works both ways doesn't mean you can stake any claim that every option is viable, or anywhere near equally likely.
As far as the Kami trio, their coverage options are pretty laughable compared to Genesect. torny and Thundy have to rely on Focus miss, and while lando gets edgequake, it's only barely above average speed for the tier. If you go for the stronger Landorus-T, you're much slower. mega Manectric also doesn't get any coverage moves other than fire moves and Hidden Power. Meanwhile, Genesect's fire/ice/electric/bug hits 10 of the 18 pure types super effectively (including relatively common water, dragon, steel, grass, psychic and dark types) and everything bar Rotom-H for neutral damage, and you seem to be forgetting that scarf genesect is a revenge killer first and a pivot second. In this aspect, comparing the kami trio is irrelevent because they are too slow to revenge kill reliably.
I understood what your calcs were trying to get across and saw something else relevent you were either ignoring, or hadn't noticed. All competitive play is about risk VS reward. The risks of leaving a Latios in on a Genesect are enormous, which makes the risk of flamethrowering against it in case the Ferrothorn comes in relatively small. Especially so if the Genesect hasn't given any indication of wanting to do anything but U-Turn. In this scenario, one of genesect's moves is always safe (relatively), U-turn. For the other player, however, one move will always result in one poke either being KO'd or heavily dented, while the other varies between scoring some chip damage on both sides or actually taking down the genesect. Like I said before: 25% chance of killing it, 50% chance of losing something, 25% chance of giving it momentum. there's only a 1 in 4 chance of the Genesect user not coming out ahead. Are you willing to risk 3 mons to get Genesect?
But all that is irrelevant if you have a hard counter. Ie, something that can safely switch into any of it's moves. Pivots are not only capable of giving team mates switch ins, the best have deadly STAB and/or coverage moves. If you think just because Torny-t is being used as a pivot means you don't need something that can absorb Heat Wave, Hurricane and Focus Blast, you've got a rude awakening coming. Having one of the four hard counters to Genesect is the only way to Guarentee it can't kill or cripple one of your team members. Some things have no counters, but these things tend to be easily checked (like Hydriegon). Something with no checks or counters is unbeatable by the definitions of checks and counters. to say pivots don't have them is silly.
I reiterate: Genesect is a revenge killer. That is it's job, that is the foremost thing it does for it's team. It should not be U-Turning against a dangerous poke that has already KO'd one of your own unless the Poke is bug weak and low on health. Losing momentum is worth a chance to do it's job. If you called a plumber because the toilet was broken, you wouldn't want them immediately looking at your sink.
Lack of STAB and low base power aren't a problem; Genesect is not a bludgeon like chomp and d-nite. It is a tactical strike, like the cut of a scalpel. It looks to revenge kill weakened offensive things and surprise kill incoming walls weak to it's attacks. Oftentimes the correct boost is irrelevent. If you expect a certain switch-in and need the special attack boost to 2hko but don't get it, it's not like it's a problem. Just use the boosted U-Turn and try to bring Genesect in on something else later.
The thing that worries me is not Genesect's momentum building or revenge killing capability; it's the combination of it's inherent prediction advantage and lack of really solid, Viable counters. the only real ones I know of (as I said before) are Heatran, Entei, Rotom-H and Megazard X.
Well, at least IMO, everything related to revenge killing is an entirely irrelevant argument when......, well, you have to die something to achieve it. You are already at a disadvantage position if you decides to do so = =
Also, the "3/4 times you are losing" is a probably a very biased argument, especially the "50% of the time you are losing something" one. I am interpreting the results of my calculation in a very different way. All those 3HKOs and 4HKOs, in my eye, means that if the coverage move is not used correctly, you actually give away free turns, I think in most scenario people should be very happy trading 30%-40% health with a turn for set ups, EH, statuses, etc.
Back to the example of Ferrothorn, if you are not doing particularly well with Ferrothorn with the rest of your team and is very unlikely to use U-turn, this only makes your choice more obvious. And your opponent can safely switch in something resisting the flamethrower predicting you doing so. Of course you can read it and U-turn instead, but that is where the mind game begins. And you are still not "risk-free" while doing so.
Concerning hard counter, you don't need a hard counter is where my argument begins. You can know if it is choice-locked through team comp analysis. What you want to do is to manage your weaknesses(which everyone should), win or at least not lose the prediction game if it comes, which is not entirely unfair if we actually weight the risks and reward on your side. Note that the opponent Genesect deserves the mini-game to be slightly inclined to its side due to occupying team slot and itself being a relatively difficult switch in.