fix(server): Improve reverse geocoded location metadata

[hermesespinola]

Description

• Limit state field to admin1Name (1st level country divisions).
• Add filter to remove 'PPLX' Feature codes from cities, which is classified as "section of populated place" (see https://www.geonames.org/export/codes.html).
• Updated tests to reflect these changes.

• Improves the reverse geocoding features. There are still edge cases.

Fixes #8941

How Has This Been Tested?

• Download photos from my Immich server.
• Upload theses photos to the dev version.
• Compare the difference between the two.

Screenshots (if appropriate):

After:

Before:

Checklist:

• I have performed a self-review of my own code
• I have made corresponding changes to the documentation if applicable

[alextran1502]

@zackpollard Feel free to merge this one after you have taken a look

[zackpollard]

I like this change for the immediate effect, however in future would it be more beneficial to keep these in but process them differently, so have more fields for storing this geo data at a lower level than city?

[hermesespinola]

What would be the use case for this data? I could create a migration to add this field to the geodata entities and filter them during reverse geocoding.

One of the main problems is that sometimes a pplx entity is close to the media location (i.e.: the San Francisco photo) even if that photo is not taken in that place (I guess because the reverse geocoding is based on proximity rather than geofencing).

[zackpollard]

What would be the use case for this data? I could create a migration to add this field to the geodata entities and filter them during reverse geocoding.

One of the main problems is that sometimes a pplx entity is close to the media location (i.e.: the San Francisco photo) even if that photo is not taken in that place (I guess because the reverse geocoding is based on proximity rather than geofencing).

Yea, perhaps the real problem is that we don't have enough of that type of data. I say we merge this and revisit it once we look at importing the entire allCountries dataset which includes way more of this finer detail data 😄