the length of time that a test plane must be out of service, i.e. approaching a year or more.
That's way too long a timescale to be credible.
If the winglet mod takes anything like a year to apply to the airframe, then no-one would buy one - it'd take a lot of savings per flight to make up for a year out-of-service.
And there's no way a test programme to clear a winglet should take anything like a year of flying - that's getting close to the time to certify a new design from scratch.
I'd have thought you could install and test a new winglet in a month or less (depending on the type); which is still a bunch of time out-of-service, but by no means prohibitive.
I'd have thought that a more credible reason is that the benefit of a winglet retrofit can't be shown to be worthwhile, because the existing tip design is sufficiently good as-is.