`crs` assignment after `read..` or creating methods for GeoDataFrame

[AndreiKingsley]

1. Add optional crs parameter to read.. methods for GeoDataFrame
   Need some research here, is it valid or not in general.
   The Shapefile always contains explicit CRS field, so no additional parameter is needed in readShapefile()
   The GeoJSON Specification claims that CRS always should be WGS84.
2. Add crs field in GeoDataFrame not nullable.

[AndreiKingsley]

So, some important updates after discussion of https://github.com/Kotlin/dataframe/pull/1640/files/8302a66491a8a9069d997975b1cc9f44c0fce1e4#r2622618050

And the problem is
Current GeoJSON support in GeoTools is not compliant with the GeoJSON specification (RFC 7946) and produces ambiguous and incorrect results.

Namely:

1. "crs" field is deprecated. As mentioned before,

the GeoJSON Specification claims that CRS always should be WGS84.

However, it allows to store geometries in any CRS (at your own risk / by some agreements), but AFAIU "crs" field should not be used nevertheless.

GeoTools writes and reads "crs" field. Moreover, it does it wrong! - it parses WGS84 CRS from geojson with the latitude first, however GeoJson points must contain longitude first. So, now write then read works wrong for GeoJson.

(also I don't note it anywhere but GeoDataFrame is designed to be geojson like, including longitude-first approach)

2. It doesn't work for 3D geometries (with Z axis), however it's allowed in GeoJSON spec.

In future we should add our own GeoJSON IO implementation - in fact it should be really easy, we just need to readJson() into DF and slightly modify it!

But now we can do more light temporary fixes:

1. make crs field in GeoDataFrame not null.
   • Shapefile, as well as other popular formats always contain it.
   • For GeoJson reading it should be WGS84 by default;
2. Add readGeoJson() overload with next arguments:
   • readOutdatedCrs: Boolean (with Warning deprecation) - if user still wants to get crs from geojson
3. Add possibility to override crs of the GeoDf (now you can only do it with geodf.df.toGeo(yourCrs)). Not sure how it shoud be done
   • In read..() - readGeoJson(customCrs: CRS)
   • geodf.setCrs(customCrs: CRS) but it can be confused with the existing applyCrs() (that transforms data)
4. Don't write crs into geojson - just add featureJSON.encodeFeatureCRS = false (by default at least).