I'm very senior these days. Most of my job appears to be unsticking people. But even I get stuck plenty, and while I can quickly unstick most problems the other devs face I can't solve all of them.
Getting stuck still happens. The real difference between what I face now and early in my career is that I have developed a sense for when to plow through, when someone else might know the answer, and when to route around.
Plowing through usually means it's not worth interrupting someone else. Or when the business itself, or other teams are in the way.
Asking questions because you don't know requires humility and it's best to have it even as a senior. Just ask, and do it sooner rather than later. You will learn faster than banging your head on the wall - there will be plenty of other problems to do that for.
Some problems aren't worth the pain. I could solve this, but should I? It saves the business lots of money to occasionally route around the problem.
I often talk about a type of pattern recognition with this sort of stuff.
Over time you start to develop a sort of automatic pattern recognition skill that separates problems to "I've never seen this before" and "oh, this looks very similar to x". Also it becomes much easier to know what's hard and what's easy, like you said.
You still obviously make the wrong call here and there. More often if you're in a completely new field, like the OP.
