Regardless of what you think is dumb, they have historically done roughly 6 year windows (sometimes less like Wii U -- Switch, sometimes more). They obviously think its smart. You have to plan out generational leaps years in advance, it can't be a reactive thing, otherwise they will have a bunch of their teams making new engines/games exclusive for Switch 2 and suddenly tell those teams to wait a year or two to release it from when they normally would. Switch had to come early because of Wii U's failures, which is why it was so dependant on Wii U ports early on. Nintendo will want the return on those games as soon as possible and there is no way, 4+ years into Switch's life, that Nintendo isn't working games for their future device. When they started Switch's life they probably mapped out all the games they wanted to make from their teams for this device before moving on to the next device's games.
And Nintendo doesn't want a year where they only sell 10-15m units. That stretches out the lifetime sales of Switch, which seems great to us that care about lifetime numbers, but they are much more concerned with profits each year (not generational profits). If they think they will sell a lot more units by having Switch 2 out then they will absolutely do that. Why sell 10-15 mil Switch 1's in 2023 when they could sell 15-20 mil Switch 2's + 5 mil Switchs (discounted) in 2023?
People always fixate on generational numbers on this website, but that's not how businesses work. Nintendo does't care about last year's financial report, shareholders don't either. It's all about the present year's financial outlook.
Furukawa has repeatedly stated that the historical 5-6 year cycle isn't smart, that's why Nintendo is looking to give Switch a longer lifecycle than previous consoles, because when the market allows it, why shouldn't Nintendo follow that. So far their actions have backed up these words, be it pricing of the console itself or game announcements which in the past used to wind down from year 5 onwards, but that isn't the case this time around.
Whenever the situation allows it (meaning good hardware momentum), Nintendo times releases to maximize profitability. So your idea that games are supposed to release as soon as they are completed fails to take into account that Nintendo isn't selling only software. Likewise, it's easy to reschedule work on games that are currently in development, so in case there would already be games in development for a Switch successor, their production timeline would get stretched out or put on hold due to higher priority for Switch games in the nearer future.
The reason why selling only 10-15m Switch units in a year would still be fine is because the userbase would still be very active and buy a ton of games. There's much more profit to be made from software than from hardware, hence why rushing out a successor doesn't make sense. Aside from that, Switch's sales momentum is too good to fall to only 10-15m in 2023.