Do not ask if there is an _opposite_ cycling lane for roundabouts

[gsurrel]

StreetComplete asks for cycling lanes in the two directions for one-way streets (which is good as some streets accepts cyclists in the two ways). However, I'm not sure if there is any roundabout that can be taken in the opposite way with a bike.

How to Reproduce

Expected Behavior

Versions affected

• 59.2

[westnordost]

It doesn't "ask". It's an overlay, not a quest. This is a difference. E.g. for private roads, the surface quest is never asked, but in the surface overlay, the information is still marked as missing.

That said, there was a similar issue recently and I thought I changed it so that the inner part of a roundabout is not marked as missing even if it is missing. I'll check.

[westnordost]

I thought I implemented this in #5716. But looking closely, it doesn't look like it. At least not this exact feature request.

[westnordost]

https://www.openstreetmap.org/way/1084085781

[westnordost]

Deciding whether cycleway tagging is not expected per-side is more complex that one might think, because it's not only junction=roundabout but also =circular and also things like highway=..._link or dual_carriageway=yes, plus, more importantly, which side of a road is the contra-flow-side depends both on tags (e.g. it could be junction=roundabout + oneway=-1) but also on in which country one is in (left hand traffic in United Kingdom).

So, I have been beginning to doubt if this is worth the added complexity or whether the overlay should just show the situation "as is". On the other hand, I did make the decision to regard this in #5716 (comment) and it is somewhat important to stay consistent in decisionmaking.

[westnordost]

Implemented in #5961, however, as there are no unit tests for cycleway overlay coloring, I'd feel better if another pair of eyes had a look over the changed code. (I tested it once at a roundabout in Lausanne and then in two roundabouts in Peterborough)