Let's say we have a "master" branch with say 1000 of commits over 2 years. We have a long-lived feature branch "feature1", about 2 months old (branched from master at that time), with 100 commits. We continue work on both master and feature1 and occasionally merge master into feature1 (about 5 commits, every 2 days) to keep it from drifting too far. Every time we do this, a particular merge conflict keeps happening, the same one each time (and I resolve it manually). For completeness, in this example it's a JavaScript project.
My question: Why does git not "remember" how I resolved this conflict the first time and use the same resolution for the subsequent merges?
Optional wrinkle which may assist with answer:
My intuition is that if I create a new branch from master and then merge feature1 into the result (let's call it "master+feature1"), the same merge conflict will stop occurring with subsequent merges from master to master+feature1. In practice I believe I have seen this, although I do not have evidence. Proof or explanation would be interesting, and I believe it could help explain the answer to my main question.
I am not interested in rebasing for this question. (Assume feature1 is pushed to a shared remote and rebasing is undesirable.)