It's quite possible for that workload to still be cheap relative to writing it yourself from scratch (e.g., I wouldn't think twice about vendoring git-subrepo).

That said, if incorporating upstream changes is more expensive than writing it yourself then...just fork the project and write new changes yourself. It's definitely possible that the project isn't a good fit and that you'd be better off starting from scratch than starting from an open-source base, but that dimension is orthogonal to the question of ongoing maintenance cost.

The concern you raise is a valid & important to consider. I’ve seen situations where there’s a mismatch between the update schedule of a component and the update schedule of the larger project & it creates an issue.