[BW] switch uuid fallback ids to address hashes

[k-nut]

As discussed, we need a more stable fallback identifier for those schools that do not have a Dienststellennummer in Baden-Württemberg. We used to rely on the WFS uuid but those ids would change between runs. This change adds a fallback hash based on the name, address, zip and city. I think this works well enough for our use case and is easy enough to explain in the readme (so that the hashes are not completely magical to end users).

Based on the latest data, the following entries have collisions when building a hash from those values. I think they are legitimate duplicates in the source data and it would be fine for us to drop them though.

[tordans]

I think they are legitimate duplicates in the source data and it would be fine for us to drop them though.

@k-nut not sure about the duplicates
For example those two look like different entities (with the same name)

• https://schulabgleich.osm-verkehrswende.org/bundesland/BW/schule/official-BW-UUID-64415520-91f0-433c-9037-137d220dfcdf?
• https://schulabgleich.osm-verkehrswende.org/bundesland/BW/schule/official-BW-UUID-546dbac8-dda0-41b0-a450-b08aabfec3be?

[k-nut]

Sorry, I think that was a misunderstanding. Those schools would still show up as two separate entities since they only share the name. Only schools with the same name, address, zip and city would be considered identical.

[tordans]

Ah, aber dann verstehe ich den Screenshot oben nicht :-)

Diese drei mit "Farbe" im Namen https://schulabgleich.osm-verkehrswende.org/bundesland/BW?map=7.72/48.59223/10.20914&lq=farbe (dort nur in der Liste)

• https://schulabgleich.osm-verkehrswende.org/bundesland/BW/schule/official-BW-UUID-7fa1dd77-493a-49da-80f8-4801662f1dd3?lq=farbe&map=13.36/48.81042/9.16653
• https://schulabgleich.osm-verkehrswende.org/bundesland/BW/schule/official-BW-UUID-fe8a204d-b782-43c3-bf6a-274bf17c2e55?lq=farbe&map=13.95/48.81178/9.17295
• https://schulabgleich.osm-verkehrswende.org/bundesland/BW/schule/official-BW-04101515?lq=farbe&map=12.52/48.81579/9.16961

Die ersten zwei wären dann die gleiche Schule in den Daten, richtig? Können wir sicherstellen, dass immer die gleiche Schule "gewinnt"?

In meinem Ansatz bleiben diese eindeutig, weil ich die gerundeten Geokoordinaten mit einbeziehe. Vermutlich overkill.

Ich würde sagen: Das hier ist auf jeden Fall besser, als das alte System, also gerne mergen von meiner Seite.