Earlier quoted context omitted.
If you say "wait 1 day without using a calendar+locale" then the duration is unambiguously 86400s, but if you say "wait 1 day using a calendar+locale" or "wait until this time tomorrow" then the duration is ambiguous until you've incorporated rules like leap/DST. I think GP's point is that "wait 1 day" unambiguously defaults to the former, and you disagree, but perhaps it's a reasonable default.
Yep, this is exactly my point. Durations are abstract spans of "stopwatch time," they don't adhere to local times or anything else we use as humans to make time more useful to us. In that context there's no real ambiguity to using units like hours/days/weeks (but not months, etc.) because they have unambiguous durations.
The collective human decision to make days-per-year vary (requiring leap rules to calculate days) seems similar to the collective human decision to make days-per-month vary (requiring month names to calculate days). So if we say a "stopwatch year" suffers the same fate as the "stopwatch month" then it's a slippery slope to saying the "stopwatch minute" is no different than a "stopwatch year" (requiring leap rules to calculate seconds) even if, for all practical purposes, it seems exempt.
I guess this is why we make "second" the SI unit, and none of our human-convenience rules mess with the duration of a second. A leap second changes the duration of a minute (and above), and a leap year changes the duration of a month (and above). Which, oddly enough, demonstrates an inconsistency: we ought to say "leap day" instead of "leap year" if the thing being added will follow the word "leap."