[GH-3161] Fix RT90 transform accuracy

[jiayuasu]

Did you read the Contributor Guide?

• Yes, I have read the Contributor Rules and Contributor Development Guide.

Is this PR related to a ticket?

• Yes, and the PR name follows the format [GH-XXX] my subject. Closes ST_Transform is off by > 180 meters since Sedona 1.9 #3161.

What changes were proposed in this PR?

Upgrade proj4sedona from 0.1.2 to 0.1.3 and add a regression test for transforming RT90 (EPSG:3021) to SWEREF99 TM (EPSG:3006).

Sedona 1.9 switched CRS transformation to proj4sedona, but version 0.1.2 lacked the generated RT90 datum definition. The transform therefore omitted the required datum shift and produced an error exceeding 180 metres. Version 0.1.3 includes the complete upstream datum registry and restores the RT90 Helmert transformation.

How was this patch tested?

• mvn spotless:apply -Dscala=2.12 -Dspark=3.4 -B
• mvn -Dspark=3.4 -Dscala=2.12 -pl spark/common -am -DskipTests install -B
• mvn -Dspark=3.4 -Dscala=2.12 -pl spark/common -Dtest=__NoJavaTests__ -Dsurefire.failIfNoSpecifiedTests=false -Dsuites=org.apache.sedona.sql.CRSTransformProj4Test test -B
• CRSTransformProj4Test: 37/37 tests passed, including the ST_Transform is off by > 180 meters since Sedona 1.9 #3161 regression.
• The proj4sedona 0.1.3 JAR used by the test was verified byte-for-byte with a fresh Maven Central download (SHA-256 2df4a9477ba4609d0b4dc34196b9fd1e44bc85b609e9ac6a726b559aad85d487).

Did this PR include necessary documentation updates?

• No, this PR does not affect any public API, so no documentation update is required.

[jiayuasu]

Superseded by #3191, which bumps proj4sedona to 0.1.4 and retains the GH-3161 RT90 regression test. The 0.1.3 build in this PR still fails 16 CRS round-trip cases that are resolved by 0.1.4.