Should we normalize the OSM values as indifferently as we do?

Created by: das-g

We currently lower-case an strip almost all OSM tag values. The list of exceptions seems somewhat arbitrary.

While the stripping is probably unproblematic, I think we should lowercase only those OSM tag values that have a descrete set of valid values. (I.e., the ones that are enum-like from the tagging scheme point of view.)

Values that are (from the tagging scheme point of view) essentially free text (e.g. all in addr:*=* tags) should probably preserve their case.