Generate crossing ways from crossing nodes when the way only needs to…

[dabreegster]

… exist on one side

Closes #69

I need to do some testing more widely, but I think this ought to do the trick

[dabreegster]

Unfortunately, this heuristic creates lots of extra gunk, in cases where a crossing node already has a way, but that way isn't tagged as a crossing -- such as https://www.openstreetmap.org/way/1293977075#map=19/52.751994/13.238300.

There's already a "Missing footway=crossing" problem that detects this problem (though I may remove a restriction that was filtering this only for severances):

Maybe these are all cases where the existing footway attached to the crossing node is an endpoint only, not in the middle. I'll try that.

[tordans]

Maybe these are all cases where the existing footway attached to the crossing node is an endpoint only, not in the middle.

Agreed

There's already a "Missing footway=crossing" problem that detects this problem (though I may remove a restriction that was filtering this only for severances):

FYI I looked into migrating the code that Rapid has to sync crossing line and nodes data to iD. Still have do document my findings but this looks promising for either iD or Speedwalk. It will require some thinking about the tagging, though. Rapid did a good start but still, one needs to understand and "own" the transformation.

[dabreegster]

Still having trouble getting this to work...