That’s a pretty contrived example. The “bypass” has maxspeed=90, the city street has maxspeed=50 on top of being longer.
But even then, are you proposing that a bypass is of higher priority than a primary, even if it’s at best a secondary?
And objectively, I wouldn’t consider that road a bypass, rather a straightening. Like you say: Everything bypasses something if you look close enough.
A bypass is a road that not only deliberately avoids destinations, but takes a significant detour in order to achieve higher speeds for through traffic.
If Czechia or another country has a similar mistake which causes the worse route to be of a higher category than the better route, then yes. Otherwise, this should never happen.
I suppose there’s no need for routers favoring routes that are bypasses as highway categories (or importance, per my concept) should be doing that job.
A bypass is a road that not only deliberately avoids destinations, but takes a significant detour in order to achieve higher speeds for through traffic.
The bypass doesn’t necessarily have to be much longer, sometimes there are tunnels that help keeping the length similar.
Besides lower speed limits in the urban areas, there are also more crossings that will reduce speed, so even without any special tagging the routers should recognize the bypass as preferable
I wouldn’t call it a bypass in this case. Maybe possibly if the tunnel runs under the city, then one might say it bypasses the city (underground), but that still feels like a stretch.
In my experience of English as spoken in Ireland, it would generally be called a bypass even if it is shorter than the original road. What makes it a bypass is that it allows through traffic to avoid the town being bypassed where that was not previously possible. Usually that happens to involve a longer distance, but even when the reverse is true, it’s still a bypass.
The Ennis Bypass (part of the M18), for example, is clearly more direct than the previous route through Ennis and Clarecastle (now the R458). It’s not an OpenStreetMap peculiarity to call it a bypass - it was referred to that way in National Roads Authority documentation at the time, and 18 years later is still called that in press reports.
At this point, a bypass basically means “There used to be a route through X, but then this road named the X bypass was built, so now the route is along this road”.
It only conveys historical information with no bearing on the current state (except the name, sometimes).
Edit: I see you removed the mentioning of Routing. That resolves this issue for me.
With that, I think the proposal is fine.
The tag also says what the road’s function for a given place is which could be useful for some statistics. It also corresponds to signs that lead to a bypass. But as you and others have pointed out, it shouldn’t affect being more favorable for routing since other tags do a better job at that.