Sure, in my opinion, this would be enough to tip the scales in favor of boundary=administrative. This expectation takes the townships beyond a mere “service area” or “special district” that coincidentally lines up with a survey boundary.
If this is OsmAnd’s actual behavior, it’s very broken and we should be careful about making data modeling decisions based on it. addr:city must matter more than boundary containment when geocoding an address. This nuance applies all across the U.S., due to how our postal system is structured, but this discussion makes it sound like they’ve intentionally discarded addr:city to economize on space.
It’s more like I’m not the authority on Maine places. I’m mostly just thinking aloud in this thread and sharing what I find out as I dig into it.
Maine didn’t use the PLSS, but the surveys in Maine also designated townships. My understanding is that Maine counties have made (civil) townships out of these (survey) townships, making them less abstract than the survey townships in a PLSS state, but not quite as real as the civil townships in one of those states. That shouldn’t be a blocker for recognizing Maine’s (civil) townships as administrative boundaries.
As to the various categories of township that you listed, I think these are just naming conventions that don’t have any rhyme or reason behind them. I would put more stock in @blackboxlogic’s preference for consistently modeling all the boundaries today recognized by the state as townships.