[regional] add support for moped links in the NL

[tyrasd]

closes #484

• adds NL-specific preset for "moped links" (highway=cycleway + bicycle=no + moped=designated)
• replace deprecation rule for highway=cycleway + bicycle=no with special "replacement" preset(s) (which excludes the NL)

[tyrasd]

this works, but it's slightly questionable if it is worth the effort of having to keep around two extra presets (plus all related translations) just to make the deprecation rule for highway=cycleway + bicycle=no work outside of the NL. For reference: Currently, there exist only ~160 features (amounting to 53km) outside of the NL which have this tag combination.

[jdhoek]

What are the alternatives?

I would guess:

• Keep the deprecation, accept lots of false positives in the Netherlands
• Remove deprecation globally until more fine-grained rules are possible

[jdhoek]

Issue #484 may of course exist in other countries as well, but I don't know if this is the case.

[tyrasd]

I think the most reasonable alternative would be to

Remove deprecation globally until more fine-grained rules are possible

One could even make the argument that having the validator rule which transforms such features into a highway=path object is even counterproductive: Just upgrading the tags potentially hides inconsistencies/errors in the data. For example, https://www.openstreetmap.org/way/217901656 might actually better be tagged as highway=footway + footway=sidewalk (and need an update of the width value) when looking at a nearby mapillary picture. An iD user could have simply "upgraded" the tags to highway=path + bicycle=no (and left the width at 2.5m), leaving the map in a sub-optimal state.

[jdhoek]

I'm inclined to agree that removing this specific deprecation for now would be better.

In the Netherlands these kind of cases are often checked periodically by mappers with a specific focus on cycling (for the Netherlands, Germany, and Belgium there is even a dedicated analysis tool that catches such errors on recreational cycling networks); I can imagine this is the case in more countries. My guess is that not that many actual errors are caught by this deprecation, and leaving it out won't do the quality of the map much harm.

Thanks for looking into this.

[tyrasd]

Sorry, I had messed up something while merging this PR. 1e75779 is the final merge commit for this one.

[1ec5]

Keep the deprecation, accept lots of false positives in the Netherlands

As a side note, it sounds pretty useful to be able to define a deprecation or other validation rule that’s sensitive to the feature’s location. We’d want to hold the line against requests for validation rules that needlessly fragment OSM tagging, but a well-justified exception in one geography could then be made without having to water down validation everywhere else.