It's insane, as you very well put it. They can't get a tattoo, too permanent, but they can get their tits chopped off. Makes total sense, yeah.
You keep trying to move the goalposts, because it's makes for more shock value to talk about operations. But the reality is that operations are a last resort, and one often brought forward due to immense mental harm for patients for whom blockers have not been an option. They are examples of the entirely predictable outcome of the kind of brutally inhumane policies you've been arguing for by opposing blockers.
Think banning surgeries will work? Trans people have been dying from illegal and unregulated surgeries for decades because it's for many been preferable to staying how they are.
Denying desperate people treatment is evil.
Plus, on top of the medical risks, starting on puberty blockers will make confused kids who didn't need them more likely to get surgeries later. That's bad, and yes, it's a trade off between not unnecessarily medicalizing confused kids and "helping" "real dysphoria" cases earlier. And I'm telling you very clearly that the answer depends on how many there are of those, but you don't want to engage with the question. Because there just aren't that many "real dysphoria" cases.
And let me preempt your next bad argument: No, a doctor can't tell them apart. That's where we are now, and they get a "dysphoria stamp" after a visit. The incentives are not aligned, they just want the money.
The goal posts have always been firmly on "Don't mess with children". I've told you many times, nobody cares what adults do, sex change surgeries have been around for a long time and nobody cares. Nobody wants to ban them on adults.
But don't mess with kids.
Time marches on in the human body, whether we like it or not. Now, show me a "perma-kitten," a cat developmentally frozen in full kittenhood, behaviorally and all (something I think would be immensely popular with some crowds) and then undo it, only then will you be close to this pause button which puberty blockers cannot well emulate.