My analysis professor used to say, if you state that some part of the proof is trivial then it should be easy to just write it out. And when you have a hard time writing it out or it takes too long, perhaps it's not trivial after all.
Not all fields of programming or mathematics are as productive of fiddly gotchas: often an experienced specialist can judge rather reliably that a piece of work will be entirely routine, though time-consuming. I can only appeal to my informal anecdotal impressions for this in programming, but in mathematics I have some harder evidence: setting up machine-checkable proofs in HOL Light. The prover won't let me get away with any hand-waving or ambiguity: I really do have to dot every single i and cross every single t. In the end, doing that is routine (though often very tedious and messy --- people talk about spending a week to formalize a page of mathematics), and I am very seldom surprised by a part of a proof that I expected to be tedious instead turning out to require new insight.